EP2481816A1

EP2481816A1 - Markers to predict survival of breast cancer patients and uses thereof

Info

Publication number: EP2481816A1
Application number: EP20120153511
Authority: EP
Inventors: Maurizio Gatti; Maria Patrizia Somma; Ferdinando Di Cunto; Paolo Provero; Christian Damasco; Antonio Lembo
Original assignee: Consiglio Nazionale delle Richerche CNR
Current assignee: Consiglio Nazionale delle Richerche CNR
Priority date: 2011-02-01
Filing date: 2012-02-01
Publication date: 2012-08-01
Also published as: IT1406754B1; ITRM20110044A1; US20120197540A1

Abstract

The present invention relates to a method to predict the mortality risk of a subject (p) affected of breast cancer comprising measuring the expression level of 105 specific genes in a biological sample, obtaining the prognostic score, S(p), that indicates the expression levels of said genes in said subject (p) affected of cancer, and predicting the mortality risk of said subj ect (p) affected of cancer.

Description

Field of invention

The present invention relates to the construction of a gene expression signature that is highly predictive of survival in breast cancer.

Background art

A reliable prediction of the outcome of a breast cancer is extremely valuable information for deciding a therapeutic strategy. The analysis of gene expression profiles obtained with microarrays has allowed identification of gene sets, or genetic "signatures", that are strongly predictive of poor prognosis (see [1,2] for a recent survey). In the past few years, two types of cancer signatures have been developed commonly designated as "bottom-up" or "top-down". In top-down (or supervised) signatures, the risk-predicting genes are selected by correlating the tumor's gene expression profiles with the patients' clinical outcome. One of the most powerful top-down signatures is the so-called 70-gene signature, which includes genes regulating cell cycle, invasion, metastasis and angiogenesis [3]. This signature outperforms standard clinical and histological criteria in predicting the likelihood of distant metastases within five years [4]. Although highly predictive of cancer outcome, top-down signatures have the drawback of including different gene types, thereby preventing precise definition of the biological processes altered in the tumor.
Bottom-up (or unsupervised) signatures are developed using sets of genes thought to be involved in specific cancer-related processes and do not rely on patients' gene expression data. Examples of these signatures are the "Wound signature" that includes genes expressed in fibroblasts after serum addition with a pattern reminiscent of the wound healing process [5,6], the "Hypoxia signatures" that contains genes involved in the transcriptional response to hypoxia [7-9], and the "Proliferation signatures" that include genes expressed in actively proliferating cells [10,11]. Other bottom-up signatures are the "ES signature" [12], the proliferation, immune response and RNA splicing modules signature [13] (henceforth abbreviated as "Module signature") the "invasiveness gene signature" (IGS) [14] and the chromosomal instability signature (CIN) [15]. The "ES signature" is based on the assumption that cells with tumor-initiating capability derive from normal stem cells. This signature reflects the gene expression pattern of embryonic stem cells (ES) and includes genes that are preferentially expressed or repressed in this type of cells [12]. The "Module signature" was generated by selecting gene sets that were enriched in nine pre-existing signatures, and consists of gene modules involved in 11 different processes including the immune response, cell proliferation, RNA splicing, focal adhesion, and apoptosis [13]. The IGS signature includes genes that are differentially expressed in tumorigenic breast cancer cells compared to normal breast-epithelium cells; the 186 genes of this signature are involved in a large variety of cellular functions and processes [14]. The CIN signature has features of both top-down and bottom-up signatures; it was developed by selecting genes with variations in the expression level correlated with the overall chromosomal aneuploidy of tumor samples [15].
Tumors are characterized by frequent mitotic divisions and chromosome instability. The authors thus reasoned that genes required for mitotic cell division and genes involved in the maintenance of chromosome integrity could be used to develop a new cancer signature.
In a recent RNAi-based screen performed in Drosophila S2 cells [16], the authors of the instant invention identified 44 genes required to prevent spontaneous chromosome breakage and 98 genes that control mitotic division. Thus, considering the strong phylogenetic conservation of the mitotic process, rather than relying on functional annotation databases, the authors used the 142 Drosophila genes identified in the screen [16] to develop a new bottom-up signature that includes genes involved in cell division but not yet annotated in the literature. 108 of these 142 Drosophila genes have unambiguous human orthologs [17]. Here the authors show that these 108 human genes constitute an excellent signature to predict breast cancer outcome. This Drosophila mitotic signature, or "DM signature", has minimal overlap with pre-existing gene signatures and outperforms them in predictive power.

Description of the invention

The classification of patients with breast cancer into risk groups represents a very valuable tool for the identification of subjects who would benefit from an aggressive systemic therapy. The analysis of microarray's data allowed to generate many signatures of gene expression improving the diagnosis and allowing the risk assessment. There is also evidence that specific genes of a proliferative state would have an high predictive value within these signatures.
Thus, the authors thus constructed a gene expression signature (the DM signature) using the human orthologues of 108 Drosophila melanogaster genes required for either the maintenance of chromosome integrity (36 genes) or mitotic division (72 genes). The DM signature has minimal overlap with the extant signatures and is highly predictive of survival in 5 large breast cancer datasets. In addition, the authors show that the DM signature outperforms other widely used cancer signatures in predictive power, and performs comparably to other proliferation-based signatures. For most genes of the DM signature, an increased expression is negatively correlated with patient survival. The genes that provide the highest contribution to the predictive power of the DM signature are those involved in cytokinesis. This finding highlights cytokinesis as an important marker in breast cancer prognosis and as a possible target for antimitotic therapies.
It is therefore, an obj ect of the invention a method to predict the mortality risk of a subj ect (p) affected of breast cancer comprising:

a) measuring the expression level of the genes C15orf44, CASP7, CNOT3, CTPS, CUL4B, CWC15, DCAKD, DDB1, FRGI, MSH6, ORC5L, PCNA, PIAS1, POLA1, PRIM2, PRPF3, RAD54L, RFC2, RPA1, RRM2, SART1, SF3A3, SMC1A, TAF6, TFDP2, TK2, TPR, TYMS, WBP11, WDR46, WDR75, XAB2, XRN2, ZMYM4, MCM3, MCM7, SMC3, NCAPD2, NCAPG, SMC4, SMC2, MASTL, ORC2L, TOP2A, CDT1, BUB3, KNTC1, ZW10, ASCC3L1, CCNB1, CDC40, DHX8, KIAA1310, LSM2, PRPF31, SF3A1, SF3A2, SF3B1, SF3B2, SF3B14, SLU7, SNRPA1, SNRPE, TXNL4A, U2AF1, U2AF2, ANAPC5, ANAPC10, CDC20,, KIN, PSMC1, SFRS15. CKAP5, EIF3A, EIF3D, EIF3E, EIF31, GTF3C3, MAPRE3, NOC3L, RRP1B, TBK1, THOC2, TUBB2C, WDR82, TRRAP, TUBGCP4, TUBG2, ASPM, CENPJ, MKI671P, PPP1R8, CDC2, KIFC1, KIF11, KIF18A, AURKC, RBBP7, PLK1, ECT2, KIF23, PRC1, RACGAP1, ANLN, CIT in a biological sample, obtaining the prognostic score, S(p), that indicates the expression levels of said genes in said subj ect (p) affected of cancer, and
b) predicting the mortality risk of said subject (p) affected of cancer comparing said prognostic score, S(p), to a cut off value (cut off threshold).

Preferably the expression level of said genes is measured by means of quantitative detection of the transcript sequences selected from the group SEQ ID No 1 to SEQ ID No. 217.
Still preferably the expression level of said genes is detected by means of microarray. In a preferred embodiment the biological sample is selected from the group of: blood, tumour cell, frozen or fixed tissue sections, biopsy, biological fluid. In a still preferred embodiment the mortality risk is assigned as follows:

i) to the class "low risk" if the prognostic score, S(p), is lower than the cut off threshold, or
ii) to the class "high risk" if the prognostic score, S(p), is greater than the cut off threshold, and optionally
iii) to the class "intermediate" if the prognostic score, S(p), is between two cut off threshold values.

Still preferably the prognostic score, S(p), is calculated according to the following formula: $S (p) = \sum_{g} x (g p) z (g)$

wherein
x(g,p) is the expression level expressed in logarithmic base 2 of the probeset g in the patient p;
z(g) is the z-score of the probeset g calculated in the Pawitan dataset;
wherein the probeset g comprises a group of 217 probes, each one being specific and selective for one of the gene transcript belonging to the group of SEQ ID No. 1 to SEQ ID No. 217.
Yet preferably the z-score for each probe is the one calculated in the Pawitan database reported in table II.
It is a further object of the invention a kit to detect the transcript expression level of genes C15orf44, CASP7, CNOT3, CTPS, CUL4B, CWC15, DCAKD, DDB1, FRG1, MSH6, ORC5L, PCNA, PIAS1, POLA1, PRIM2, PRPF3, RAD54L, RFC2, RPA1, RRM2, SART1, SF3A3, SMC1A, TAF6, TFDP2, TK2, TPR, TYMS, WBP11, WDR46, WDR75, XAB2, XRN2, ZMYM4, MCM3, MCM7, SMC3, NCAPD2, NCAPG, SMC4, SMC2, MASTL, ORC2L, TOP2A, CDT1, BUB3, KNTC1, ZW10, ASCC3L1, CCNB1, CDC40, DHX8, KIAA1310, LSM2, PRPF31, SF3A1, SF3A2, SF3B1, SF3B2, SF3B14, SLU7, SNRPA1, SNRPE, TXNL4A, U2AF1, U2AF2, ANAPC5, ANAPC10, CDC20,, KIN, PSMC1, SFRS15. CKAP5, EIF3A, EIF3D, EIF3E, EIF3I, GTF3C3, MAPRE3, NOC3L, RRP1B, TBKI, THOC2, TUBB2C, WDR82, TRRAP, TUBGCP4, TUBG2, ASPM, CENPJ, MKI67IP, PPP1R8, CDC2, KIFC1, KIF11, KIF18A, AURKC, RBBP7, PLK1, ECT2, KIF23, PRC1, RACGAP1, ANLN, CIT,
comprising:

for each of said genes, sequence specific amplification means to obtain amplified nucleic acids having sequences comprised in the transcribed region thereof;
quantitative detection means of said amplified nucleic acids;
appropriate reagents.

Preferably said amplified nucleic acids consist of :

for C15orf44, SEQ ID No. 145; for CASP7, SEQ ID No. 189; for CNOT3, SEQ ID No. 66 and/or SEQ ID No. 138 and/or SEQ ID No.167; for CTPS, SEQ ID No. 39; for CUL4B, SEQ ID No. 113 and/or SEQ ID No. 152 and/or SEQ ID No. 165 and/or SEQ ID No. 212; for CWC15, SEQ ID No. 159; for DCAKD, SEQ ID No. 126 and/or SEQ ID No. 140 and/or SEQ ID No. 190; for DDB1, SEQ ID No. 38; for FRG1, SEQ ID No. 195; for MSH6, SEQ ID No. 46 and/or SEQ ID No. 61 and /or SEQ ID No. 153 and/or SEQ ID No. 187; for ORC5L, SEQ ID No. 70 and/or SEQ ID No. 79 and/or SEQ ID No. 109; for PCNA, SEQ ID No. 51; for PIAS1, SEQ ID No. 211 and/or SEQ ID No. 216 and/or SEQ ID No. 217; for POLA1, SEQ ID No. 147; for PRIM2, SEQ ID No. 43 and/or SEQ ID No. 56 and/or SEQ ID No. 88; for PRPF3, SEQ ID No. 170; for RAD54L, SEQ ID No. 75; for RFC2, SEQ ID No. 42 and/or SEQ ID No. 48; for RPA1, SEQ ID No. 64 and/or SEQ ID No. 103; for RRM2, SEQ ID No. 3 and/or SEQ ID No. 9; for SART1, SEQ ID No. 124; for SF3A3, SEQ ID No. 201; for SMC1A, SEQ ID No. 115 and/or SEQ ID No. 179 and/or SEQ ID No. 207; for TAF6, SEQ ID No. 68;
for TFDP2, SEQ ID No. 86 and/or SEQ ID No. 118 and/or SEQ ID No. 210; for TK2, SEQ ID No. 37 and/or SEQ ID No. 156 and/or SEQ ID No. 171 and/or SEQ ID No. 172; for TPR, SEQ ID No. 99 and/or SEQ ID No. 108 and/or SEQ ID No. 182 and/or SEQ ID No. 204; for TYMS, SEQ ID No. 32 and/or SEQ ID No. 125; for WBP11, SEQ ID No. 65 and/or SEQ ID No. 67; for WDR46, SEQ ID No. 93; for WDR75, SEQ ID No. 158; for XAB2, SEQ ID No. 180; for XRN2, SEQ ID No. 81 and/or SEQ ID No. 84; for ZMYM4, SEQ ID No. 192 and/or SEQ ID No. 196 and/or SEQ ID No. 213; for MCM3, SEQ ID No. 34; for MCM7, SEQ ID No. 28 and/or SEQ ID No. 52; for SMC3, SEQ ID No. 185 and/or SEQ ID No. 193 and/or SEQ ID No. 209; for NCAPD2, SEQ ID No. 106; for NCAPG, SEQ ID No.22 and/or SEQ ID No. 24; for SMC4, SEQ ID No. 33 and/or SEQ ID No. 54 and/or SEQ ID No. 141; for SMC2, SEQ ID No. 45 and/or SEQ ID No. 127; for MASTL, SEQ ID No. 11; for ORC2L, SEQ ID No. 104; for TOP2A, SEQ ID No. 20 and/or SEQ ID No. 62 and/or SEQ ID No. 96; for CDT1, SEQ ID No. 2 and/or SEQ ID No. 36; for BUB3, SEQ ID No. 57 and/or SEQ ID No. 139 and/or SEQ ID No. 148 and/or SEQ ID No. 174 and/or SEQ ID No. 178; for KNTC1, SEQ ID No. 35; for ZW10, SEQ ID No. 143; for ASCC3L1, SEQ ID No. 55 and/or SEQ ID No. 135 and/or SEQ ID No. 150; for CCNB1, SEQ ID No. 7 and/or SEQ ID No. 14; for CDC40, SEQ ID No. 100 and/or SEQ ID No. 177; for DHX8, SEQ ID No. 58 and/or SEQ ID No. 120 and/or SEQ ID No. 121; for KIAA1310, SEQ ID No. 160 and/or SEQ ID No. 183 and/or SEQ ID No. 188; for LSM2, SEQ ID No. 137; for PRPF31, SEQ ID No. 60 and/or SEQ ID No. 91 and/or SEQ ID No. 184; for SF3A1, SEQ ID No. 98 and/or SEQ ID No. 119 and/or SEQ ID No. 162 and/or SEQ ID No. 173; for SF3A2, SEQ ID No. 169 and/or SEQ ID No. 176; for SF3B1, SEQ ID No. 194 and/or SEQ ID No. 203 and/or SEQ ID No. 208 and/or SEQ ID No. 214; for SF3B2, SEQ ID No. 77; for SF3B14, SEQ ID No. 10; for SLU7, SEQ ID No. 149 and/or SEQ ID No. 151; for SNRPA1, SEQ ID No. 23 and/or SEQ ID No. 49 and/or SEQ ID No. 71 and/or SEQ ID No. 181; for SNRPE, SEQ ID No. 72 and/or SEQ ID No. 136; for TXNL4A, SEQ ID No. 26 and/or SEQ ID No. 134; for U2AF1, SEQ ID No. 30 and/or SEQ ID No. 82 and/or SEQ ID No. 102 and/or SEQ ID No. 131; for U2AF2, SEQ ID No. 94 and/or SEQ ID No. 146 and/or SEQ ID No. 155 and/or SEQ ID No. 161; for ANAPC5, SEQ ID No. 85 and/or SEQ ID No. 95 and/or SEQ ID No. 97 and/or SEQ ID No. 112 and/or SEQ ID No. 117; for ANAPC10, SEQ ID No. 129; for CDC20, SEQ ID No. 17; for KIN, SEQ ID No. 111 and/or SEQ ID No. 144; for PSMC1, SEQ ID No. 25; for SFRS15, SEQ ID No. 50 and/or SEQ ID No. 63 and/or SEQ ID No. 80 and/or SEQ ID No. 142 and/or SEQ ID No. 197; for CKAP5, SEQ ID No. 21; for EIF3A, SEQ ID No. 175 and/or SEQ ID No. 186 and/or SEQ ID No. 202; for EIF3D, SEQ ID No. 101; for EIF3E, SEQ ID No. 154; for EIF3I, SEQ ID No. 114; for GTF3C3, SEQ ID No. 74 and/or SEQ ID No. 163; for MAPRE3, SEQ ID No. 116 and/or SEQ ID No. 128 and/or SEQ ID No. 130 and/or SEQ ID No. 133; for NOC3L, SEQ ID No. 164; for RRP1B, SEQ ID No. 105 and/or SEQ ID No. 123; for TBK1, SEQ ID No. 198; for THOC2, SEQ ID No. 110 and/or SEQ ID No. 132 and/or SEQ ID No. 199 and/or SEQ ID No. 205; for TUBB2C, SEQ ID No. 4 and/or SEQ ID No. 5; for WDR82, SEQ ID No. 191; for TRRAP, SEQ ID No. 69 and/or SEQ ID No. 73; for TUBGCP4, SEQ ID No. 76 and/or SEQ ID No. 215; for TUBG2, SEQ ID No. 157; for ASPM, SEQ ID No. 6 and/or SEQ ID No. 47 and/or SEQ ID No. 53; for CENPJ, SEQ ID No. 87 and/or SEQ ID No. 92 and/or SEQ ID No. 107; for MKI67IP, SEQ ID No. 41 and/or SEQ ID No. 89 and/or SEQ ID No. 200; for PPP1R8, SEQ ID No. 168; for CDC2, SEQ ID No. 15 and/or SEQ ID No. 16 and/or SEQ ID No. 31 and/or SEQ ID No. 206; for KIFC1, SEQ ID No. 19; for KIF11, SEQ ID No. 29; for KIF18A, SEQ ID No. 18; for AURKC, SEQ ID No. 90; for RBBP7, SEQ ID No. 166; for PLK1, SEQ ID No. 27; for ECT2, SEQ ID No. 40 and/or SEQ ID No. 59 and/or SEQ ID No. 83; for KIF23, SEQ ID No. 8 and/or SEQ ID No. 44; for PRC1, SEQ ID No. 13; for RACGAP1, SEQ ID No. 12; for ANLN, SEQ ID No. 1; for CIT, SEQ ID No. 78 and/or SEQ ID No. 122.

Still preferably, the kit further comprises sequence specific amplification means to obtain amplified nucleic acids having sequences comprised in the transcribed region of genes H3F3A and/or PPAN-P2RY11 and/or KIF4.
It is a further object of the invention a microarray consisting of:

a) solid supporting means, and
b) for each of the genes C15orf44, CASP7, CNOT3, CTPS, CUL4B, CWC15, DCAKD, DDB1, FRG1, MSH6, ORC5L, PCNA, PIAS1, POLA1, PRIM2, PRPF3, RAD54L, RFC2, RPA1, RRM2, SART1, SF3A3, SMC1A, TAF6, TFDP2, TK2, TPR, TYMS, WBP11, WDR46, WDR75, XAB2, XRN2, ZMYM4, MCM3, MCM7, SMC3, NCAPD2, NCAPG, SMC4, SMC2, MASTL, ORC2L, TOP2A, CDT1, BUB3, KNTC1, ZW10, ASCC3L1, CCNB1, CDC40, DHX8, KIAA1310, LSM2, PRPF31, SF3A1, SF3A2, SF3B1, SF3B2, SF3B14, SLU7, SNRPA1, SNRPE, TXNL4A, U2AF1, U2AF2, ANAPC5, ANAPC10, CDC20, KIN, PSMC1, SFRS15. CKAP5, EIF3A, EIF3D, EIF3E, EIF3I, GTF3C3, MAPRE3, NOC3L, RRP1B, TBKI, THOC2, TUBB2C, WDR82, TRRAP, TUBGCP4, TUBG2, ASPM, CENPJ, MKI67IP, PPP1R8, CDC2, KIFC1, KIF11, KJF18A, AURKC, RBBP7, PLK1, ECT2, KIF23, PRC1, RACGAP1, ANLN, CIT, at least one oligonucleotide able to specifically hybridize to a sequence comprised in the transcribed region thereof. Preferably wherein the sequences comprised in the transcribed region of said genes consist of: for C15orf44, SEQ ID No. 145; for CASP7, SEQ ID No. 189; for CNOT3, SEQ ID No. 66 and/or SEQ ID No. 138 and/or SEQ ID No.167; for CTPS, SEQ ID No. 39; for CUL4B, SEQ ID No. 113 and/or SEQ ID No. 152 and/or SEQ ID No. 165 and/or SEQ ID No. 212; for CWC15, SEQ ID No. 159; for DCAKD, SEQ ID No. 126 and/or SEQ ID No. 140 and/or SEQ ID No. 190; for DDB1, SEQ ID No. 38; for FRG1, SEQ ID No. 195; for MSH6, SEQ ID No. 46 and/or SEQ ID No. 61 and /or SEQ ID No. 153 and/or SEQ ID No. 187; for ORC5L, SEQ ID No. 70 and/or SEQ ID No. 79 and/or SEQ ID No. 109; for PCNA, SEQ ID No. 51; for PIAS1, SEQ ID No. 211 and/or SEQ ID No. 216 and/or SEQ ID No. 217; for POLA1, SEQ ID No. 147; for PRIM2, SEQ ID No. 43 and/or SEQ ID No. 56 and/or SEQ ID No. 88; for PRPF3, SEQ ID No. 170; for RAD54L, SEQ ID No. 75; for RFC2, SEQ ID No. 42 and/or SEQ ID No. 48; for RPA1, SEQ ID No. 64 and/or SEQ ID No. 103; for RRM2, SEQ ID No. 3 and/or SEQ ID No. 9; for SART1, SEQ ID No. 124; for SF3A3, SEQ ID No. 201; for SMC1A, SEQ ID No. 115 and/or SEQ ID No. 179 and/or SEQ ID No. 207; for TAF6, SEQ ID No. 68; for TFDP2, SEQ ID No. 86 and/or SEQ ID No. 118 and/or SEQ ID No. 210; for TK2, SEQ ID No. 37 and/or SEQ ID No. 156 and/or SEQ ID No. 171 and/or SEQ ID No. 172; for TPR, SEQ ID No. 99 and/or SEQ ID No. 108 and/or SEQ ID No. 182 and/or SEQ ID No. 204; for TYMS, SEQ ID No. 32 and/or SEQ ID No. 125; for WBP11, SEQ ID No. 65 and/or SEQ ID No. 67; for WDR46, SEQ ID No. 93; for WDR75, SEQ ID No. 158; for XAB2, SEQ ID No. 180; for XRN2, SEQ ID No. 81 and/or SEQ ID No. 84; for ZMYM4, SEQ ID No. 192 and/or SEQ ID No. 196 and/or SEQ ID No. 213; for MCM3, SEQ ID No. 34; for MCM7, SEQ ID No. 28 and/or SEQ ID No. 52; for SMC3, SEQ ID No. 185 and/or SEQ ID No. 193 and/or SEQ ID No. 209; for NCAPD2, SEQ ID No. 106; for NCAPG, SEQ ID No.22 and/or SEQ ID No. 24; for SMC4, SEQ ID No. 33 and/or SEQ ID No. 54 and/or SEQ ID No. 141; for SMC2, SEQ ID No. 45 and/or SEQ ID No. 127; for MASTL, SEQ ID No. 11; for ORC2L, SEQ ID No. 104; for TOP2A, SEQ ID No. 20 and/or SEQ ID No. 62 and/or SEQ ID No. 96; for CDT1, SEQ ID No. 2 and/or SEQ ID No. 36; for BUB3, SEQ ID No. 57 and/or SEQ ID No. 139 and/or SEQ ID No. 148 and/or SEQ ID No. 174 and/or SEQ ID No. 178; for KNTC1, SEQ ID No. 35; for ZW10, SEQ ID No. 143; for ASCC3L1, SEQ ID No. 55 and/or SEQ ID No. 135 and/or SEQ ID No. 150; for CCNB1, SEQ ID No. 7 and/or SEQ ID No. 14; for CDC40, SEQ ID No. 100 and/or SEQ ID No. 177; for DHX8, SEQ ID No. 58 and/or SEQ ID No. 120 and/or SEQ ID No. 121; for KIAA1310, SEQ ID No. 160 and/or SEQ ID No. 183 and/or SEQ ID No. 188; for LSM2, SEQ ID No. 137; for PRPF31, SEQ ID No. 60 and/or SEQ ID No. 91 and/or SEQ ID No. 184; for SF3A1, SEQ ID No. 98 and/or SEQ ID No. 119 and/or SEQ ID No. 162 and/or SEQ ID No. 173; for SF3A2, SEQ ID No. 169 and/or SEQ ID No. 176; for SF3B1, SEQ ID No. 194 and/or SEQ ID No. 203 and/or SEQ ID No. 208 and/or SEQ ID No. 214; for SF3B2, SEQ ID No. 77; for SF3B14, SEQ ID No. 10; for SLU7, SEQ ID No. 149 and/or SEQ ID No. 151; for SNRPA1, SEQ ID No. 23 and/or SEQ ID No. 49 and/or SEQ ID No. 71 and/or SEQ ID No. 181; for SNRPE, SEQ ID No. 72 and/or SEQ ID No. 136; for TXNL4A, SEQ ID No. 26 and/or SEQ ID No. 134; for U2AF1, SEQ ID No. 30 and/or SEQ ID No. 82 and/or SEQ ID No. 102 and/or SEQ ID No. 131; for U2AF2, SEQ ID No. 94 and/or SEQ ID No. 146 and/or SEQ ID No. 155 and/or SEQ ID No. 161; for ANAPC5, SEQ ID No. 85 and/or SEQ ID No. 95 and/or SEQ ID No. 97 and/or SEQ ID No. 112 and/or SEQ ID No. 117; for ANAPC10, SEQ ID No. 129; for CDC20, SEQ ID No. 17; for KIN, SEQ ID No. 111 and/or SEQ ID No. 144; for PSMC1, SEQ ID No. 25; for SFRS15, SEQ ID No. 50 and/or SEQ ID No. 63 and/or SEQ ID No. 80 and/or SEQ ID No. 142 and/or SEQ ID No. 197; for CKAP5, SEQ ID No. 21; for EIF3A, SEQ ID No. 175 and/or SEQ ID No. 186 and/or SEQ ID No. 202; for EIF3D, SEQ ID No. 101; for EIF3E, SEQ ID No. 154; for EIF3I, SEQ ID No. 114; for GTF3C3, SEQ ID No. 74 and/or SEQ ID No. 163; for MAPRE3, SEQ ID No. 116 and/or SEQ ID No. 128 and/or SEQ ID No. 130 and/or SEQ ID No. 133; for NOC3L, SEQ ID No. 164; for RRP1B, SEQ ID No. 105 and/or SEQ ID No. 123; for TBK1, SEQ ID No. 198; for THOC2, SEQ ID No. 110 and/or SEQ ID No. 132 and/or SEQ ID No. 199 and/or SEQ ID No. 205; for TUBB2C, SEQ ID No. 4 and/or SEQ ID No. 5; for WDR82, SEQ ID No. 191; for TRRAP, SEQ ID No. 69 and/or SEQ ID No. 73; for TUBGCP4, SEQ ID No. 76 and/or SEQ ID No. 215; for TUBG2, SEQ ID No. 157; for ASPM, SEQ ID No. 6 and/or SEQ ID No. 47 and/or SEQ ID No. 53; for CENPJ, SEQ ID No. 87 and/or SEQ ID No. 92 and/or SEQ ID No. 107; for MKI67IP, SEQ ID No. 41 and/or SEQ ID No. 89 and/or SEQ ID No. 200; for PPP1R8, SEQ ID No. 168; for CDC2, SEQ ID No. 15 and/or SEQ ID No. 16 and/or SEQ ID No. 31 and/or SEQ ID No. 206; for KIFC1, SEQ ID No. 19; for KIF11, SEQ ID No. 29; for KIF18A, SEQ ID No. 18; for AURKC, SEQ ID No. 90; for RBBP7, SEQ ID No. 166; for PLK1, SEQ ID No. 27; for ECT2, SEQ ID No. 40 and/or SEQ ID No. 59 and/or SEQ ID No. 83; for KIF23, SEQ ID No. 8 and/or SEQ ID No. 44; for PRC1, SEQ ID No. 13; for RACGAP1, SEQ ID No. 12; for ANLN, SEQ ID No. 1; for CIT, SEQ ID No. 78 and/or SEQ ID No. 122.

Preferably the microarray further comprises at least one oligonucleotide able to specifically hybridize to a sequence comprised in the transcribed region of genes H3F3A and/or PPAN-P2RY11 and/or KIF4.
In the present invention the method to predict the mortality risk of a subject affected of breast cancer is also a method to predict the survival of a subject affected of breast cancer. Further the genes of the DM signature could be merged with those of other signatures to further improve risk stratification.
In the present invention, 3 cutoff values are provided, corresponding to 90%, 70% and 50% sensitivity on Miller dataset.
The cut off threshold on the prognostic score were calculated on the Miller dataset (a dataset independent from that used to develop the signature, but built on a consecutive series of patients and therefore representative of the population), and corresponds, on this dataset, to 90%, 70% and 50% sensitivity. Sensitivity is defined as the fraction of high-risk patients correctly identified by the predictor. For each cut off, the specificity is reported. The specificity was calculated on the Miller dataset and is defined as the fraction of low-risk patients correctly identified by the predictor. The cut off of 90% sensitivity = 798 (32% specificity), the cut off of 70% sensitivity = 921.8 (57% specificity) and the cut off of 50% sensitivity = 928.5 (73% specificity). These values are non-limitative example and may vary.
The present invention is illustrated by the following non limiting examples and figures.

Fig. 1 - Predictive power of the DM signature. Kaplan-Meier analysis using the DM signature shows significant differences in survival of patients from five independents breast cancer datasets. The curves represent the cumulative chances of survival of patients classified within two groups by the hierarchic clustering algorithm based on the correlation coefficient: lower curve - high risk patients; top curve - low risk patients.
Fig. 2 - Predictive power of the mitotic and chromosome-integrity genes of the DM signature. Kaplan-Meier survival analysis was performed on five breast cancer datasets using either the 34 chromosome integrity genes or the 71 mitotic genes of the DM signature represented in the Affymetrix platform. The curves represent the cumulative probabilities of survival of patients classified within two groups by the hierarchic clustering algorithm based on the correlation coefficient: lower curve - high risk patients; top curve - low risk patients.
Fig. 3 - The DM signature outperforms 9 major signatures in predictive power. The predictive power of signature is expressed with P; P is the P-value of the log-rank test for difference in survival probability of the two groups of patients obtained by hierarchical clustering using the genes of each signature. Colours correspond to the statistical significance: red, P >= 0.05; yellow, 0.05>P >= 0.01; green, P<0.01. The signatures compared (DM; Proliferation of Starmans et al. [11], Module [13], CIN [15], Hypoxia of Sung et al. [8], Hypoxia of Winter et al. [9], ES [12]; 70-gene [3]; IGS [14]; Wound [5,6] are described in the text.
Fig. 4 - Distribution of the z-scores of the genes of the DM signature compared to the distribution of z-scores of all genes represented in five breast cancer datasets. Density = ratio between the number of the genes in a given z-score and the total number of genes.
Fig. 5 - Comparative evaluation of the prognostic score of the DM signature. The prognostic score of the DM signature is compared to those obtained from the CIN [15], Proliferation [11], IGS [14], Hypoxia [9], 70-gene [3], and Wound [5] signatures in the three datasets not used for training. The scores are used to predict outcome at five years. The bars show the areas under the ROC curves (AUC).
Fig. 6 - Predictive power of the DM signature on a dataset of lung cancer [18]: Kaplan-Meier survival analysis. The curves represent the cumulative probabilities of survival of patients classified within two groups by the hierarchic clustering algorithm based on the correlation coefficient: lower curve - high risk patients; top curve - low risk patients.
Fig. 7 - Predictive power of the DM signature on a dataset of glioma [19]: Kaplan-Meier survival analysis. The curves represent the cumulative probabilities of survival of patients classified within two groups by the hierarchic clustering algorithm based on the correlation coefficient: lower curve - high risk patients; top curve - low risk patients.

Materials and methods

Definition of the DM signature

The 142 D. melanogaster mitotic genes described in [16] were first converted into Entrez gene ids (file gene_info.gz downloaded from the Entrez Gene ftp site in June 2008). The authors then used Homologene, build 62, to obtain the 108 human orthologues that compose the DM signature. The authors considered only one-to-one orthology relationships reported in Homologene. This criterion led to the exclusion from the DM signature of several human genes that are commonly considered homologous to the Drosophila genes. However, the degree of homology between these human genes and their Drosophila counterparts was not sufficient for inclusion in Homologene.

Breast cancer datasets

The authors used the following publicly available breast cancer datasets: NKI [4]; Pawitan [20] - Gene Expression Omnibus (GEO-) series GSE1456; Miller [21] - GEO series GSE3494; Wang [22] - GEO series GSE2034; Desmedt [23] - GEO series GSE7390; and Sotiriou [24] - GEO series GSE2990. The authors used relapse-free survival times when available, and overall survival times otherwise. Since the Sotiriou, Desmedt and Miller datasets have some patients in common, the authors merged the Sotiriou and Desmedt datasets in a single dataset, from which the authors removed the patients included in the Miller dataset. The authors refer to this combined dataset as the Sotiriou-Desmedt dataset. Normalized expression data and clinical data for the NKI dataset were obtained from http://www.rii.com/publications/2002/nejm.html. For the Affymetrix-based datasets, the authors obtained gene expression values from the raw data, using MAS 5.0 algorithm as implemented in the simpleaffy [25] package of Bioconductor [26]. For all datasets the authors considered only the probesets unambiguously assigned to one Entrez Gene ID in the platform annotation. For the Affymetrix platform, the authors used the annotation provided by the manufacturer, version 25, which allowed them to identify single or multiple probesets for 105 of the 108 DM signature genes. For the NKI dataset the authors used the annotation file provided in the website mentioned above; the correspondence between sequence accession number and Entrez gene was obtained from the Entrez gene ftp site; 98 of the 108 DM genes were thus associated with one or multiple probes.

Dataset of patients with lung glandular cancer and of patients with glioma.

The expression data of patient with lung glandular cancer [18] were obtained from the caArray database, (https://array.nci.nih.gov/caarray) identification "jacobs-00182". The expression data of patients with glioma [19] were obtained by the GEO database, accession GSE4271. In both cases data were treated as described for the breast cancer dataset on Affymetrix platform.
The Large lung cancer dataset refers to bibliographic reference [18]. Other lung cancer dataset and also ovarian cancer refer to bibliographic reference [27].

Determination of the predictive power of the genes in the DM signatures by clustering analysis

To determine whether the expression profiles of the genes included in the DM signature are significantly and robustly correlated with the disease outcome the authors used the following procedure on the datasets mentioned above: (a) select the microarray probes unambiguously associated to the signature genes; (b) creating two groups of patients by Pearson correlation-based hierarchical clustering, using only the expression profiles of the probes selected in step a; (c) determining by a standard log-rank test, as implemented in the survival library of R, whether the cumulative probability of survival is significantly different between the two groups.

Determination of prognostic scores

For all datasets the authors divided the patients into two groups (good- and poor-outcome) based on their status at five years. The authors then calculated the prognostic scores for outcome prediction at five years using the following procedures. For the 70-gene signature, the score of a patient is the cosine-correlation of the expression profile of genes with good-prognosis found in http://www.rii.com/publications/2002/nejm.html [4]. The genes in the signature, given at as accession numbers, were translated into Entrez gene IDs and then into Affymetrix probesets using Affymetrix annotation files, version 25. The authors obtained 76 probesets for the HG-U133A platform, and 109 probesets for the HG-U133A and HG-U133B platforms considered together. Probesets corresponding to the same gene were assigned the same coefficient in the good-prognosis profile.
For the Wound and IGS signatures, the score of a patient is given by the Pearson correlation of the expression profile of the signature genes. For the Wound signature the core serum response centroid is available at http://microarray-pubs.stanford.edu/wound [5]. The genes in the signature were translated into Entrez gene ids and then into Affymetrix probesets using the procedure described above. The authors obtained 493 probesets for the HG-U133A platform, and 667 probesets for the HG-U133A and HG-U133B platforms considered together. Probesets corresponding to the same gene were assigned the same expression value in the core serum response centroid. The centroid for the IGS signature is directly given in Affymetrix probesets [14].
For the CIN [15], Proliferation [11] and Hypoxia [9] signatures, the score of a patient is the sum of the logarithmic expression of the signature genes in the patient sample. For the CIN and Proliferation signatures, the gene symbols, were translated first into Entrez gene ids and then into Affymetrix probesets as described above. The Hypoxia signature is directly given in terms of Affymetrix probesets.
For the DM signature, the prognostic score of a patient is given by: $S (p) = \sum_{g} x (g p) z (g)$

where the sum is over all the probesets associated to the signature, z(g) is the z-score of probeset g computed in the Pawitan dataset and x(g,p) is the logarithmic expression level of probeset g in patient p. The Affymetrix probesets that comprise the DM signature together with their z-scores are reported in Table II.
The authors used ROC curves to compare the scalable scores on three datasets (Miller, Wang and Sotiriou-Desmedet). The area under the curves and the related standard error were computed using the Hmisc library and programs available at http://biostat.mc.vanderbilt.edu/s/Hmisc. The Pawitan and NKI datasets were not used in this comparison because they were involved in the training of the DM and 70-gene signatures, respectively.

Contribution of specific gene classes to the predictive power of the signature

The contribution of each probeset g to the difference in score between poor- and good-prognosis patients is defined as: $Δ s (g) = z (g) (P (g) - G (g))$

where P(g) (G(g)) is the logarithmic expression of the probeset averaged on all poor (good) prognosis patients and z(g) is the z-score of the probeset. Given a subset of the DM signature (e.g. cytokinesis-related genes), the authors used a Mann-Whitney U test to compare the contribution of the probesets included in the subset to the contribution of all the other probesets.

mRNA amplification

The methods for obtaining and amplifying mRNA are known in the art and described for example in Sambrook et al., Molecular Cloning - A laboratory manual (2nd Ed.), vols. 1-3, Cold Spring Harbor Laboratory, Cold Spring Harbor, New York (1989) and Ausubel et al. Current Protocols in Molecular Biology vol.2, Current Protocol Publishing, New York (1994).
The RNA can be isolated from samples of tumor tissue, frozen or fixed tumor tissue sections, biopsy, biological fluid or tumor cell.
In the method, the sequence can be in any part of the transcript as indicated in Table II.

Results

Generation of the DM signature

The authors have recently carried out an RNAi-based screen to detect Drosophila genes required for chromosome integrity and for the fidelity of mitotic division [16]. Since these types of genes tend to be transcriptionally co-expressed, the authors first used a co-expression-based bioinformatic procedure to select a group of 1,000 genes highly enriched in mitotic functions. The authors then performed RNAi against each of these genes in Drosophila S2 cultured cells. Phenotypic analysis of dsRNA-treated cells allowed the identification of 142 genes representative of the entire spectrum of functions required for proper transmission of genetic information. 44 of these genes were required to prevent spontaneous chromosome breakage. The remaining 98 genes specified a variety of mitotic functions including those required for spindle assembly, chromosome segregation and cytokinesis [16]. Based on the observed RNAi phenotypes, these 142 genes were subdivided into 18 phenoclusters [16].

To construct the DM signature the authors identified the human homologues of these Drosophila genes, according to Homologene [17]. Both the genes required for chromosome integrity and those involved in the mitotic process turned out to be highly conserved in humans. 36 of the 44 chromosome-integrity genes and 72 of the 98 mitotic genes had clear human orthologues. These 108 human genes, and their classification according to the phenotypes associated with RNAi-mediated silencing of their Drosophila counterparts, are listed in Tables I and II.

Table I. Classification of the 108 genes of the DM signature according to the RNAi phenotypes of their Drosophila orthologues. The phenoclusters, indicated in bold characters, are described in detail in [16]

RNAi phenotypes elicited by the Drosophila genes	Names of the human orthologues
Chromosome integrity genes
Chromosome aberrations (CA)	C15orf44, CASP7, CNOT3, CTPS, CUL4B, CWC15, DCAKD, DDB1, FRGI, H3F3A, MSH6, ORC5L, PCNA, PIAS1, PPAN-P2RY11, POLA1, PRIM2, PRPF3, RAD54L, RFC2, RPA1, RRM2, SART1, SF3A3, SMC1A, TAF6, TFDP2, TK2, TPR, TYMS, WBP11,
	WDR46, WDR75, XAB2, XRN2, ZMYM4.

Mitotic genes
Abnormal chromosome structure. CC1, loss of sister chromatid cohesion in heterochromatin; CC2 and CC3, defective lateral and longitudinal chromosome condensation, respectively	CC1: MCM3, MCM7, SMC3.
	CC2: NCAPD2, NCAPG, SMC4, SMC2.
	CC3: MASTL, ORC2L, TOP2A.
Abnormal chromosome segregation. CS1, defective chromosome duplication; CS2, precocious sister chromatid separation; CS3 and CS4, lack of sister chromatid separation; CS5, defective chromosome segregation during anaphase	CS1: CDT1.
	CS2: BUB3, KNTC1, ZW10.
	CS3 and CS4: ASCC3L1, CCNB1, CDC40, DHX8, KIM1310, LSM2, PRPF31, SF3A1, SF3A2, SF3B1, SF3B2, SF3B14, SLU7, SNRPA1, SNRPE, TXNL4A, U2AF1, U2AF2.
	CS5: ANAPC5, ANAPC10, CDC20, KIF4A, KIN, PSMC1, SFRS15.
Abnormal spindle morphology: SA1, short spindles; SA2, spindles with a low MT density; SA3, poorly focused spindle poles, SA4 miscellaneous spindle defects	SA1: CKAP5, EIF3A, EIF3D, EIF3E, EIF3I, GTF3C3, MAPRE3, NOC3L, RRP1B, TBK1, THOC2, TUBB2C, WDR82.
	SA2: TRRAP, TUBGCP4, TUBG2.
	SA3: ASPM, CENPJ, MKI67IP, PPP1R8.
	SA4: CDC2, KIFC1, KIF11, KIF18A.
Abnormal spindle and chromosome structure: SC1, defective chromosome condensation and cytokinesis; SC2, multiple mitotic defects	SC1: AURKC, RBBP7.
	SC2: PLK1.
Frequent cytokinesis failures: CY1 and CY2, defective in early and late cytokinesis, respectively	CY1: ECT2, KIF23, PRC1, RACGAP1.
	CY2: ANLN, CIT.

Table II - Ranking of the Affymetrix probesets of the DM signature according to their z-scores. The Affymetrix probesets associated with the DM signature genes are ranked according to their Cox z-score computed on the training dataset (Pawitan). The contribution to the difference in score between poor and good prognosis patients in the other datesets is also reported. The phenoclusters associated with the Drosophila genes [16] are abbreviated as follows: CA, chromosome aberrations; CC1, loss of sister chromatid cohesion in heterochromatin; CC2 aberrant lateral chromosome condensation; CC3, aberrant longitudinal chromosome condensation; CS1, defective chromosome duplication; CS2, precocious sister chromatid separation; CS3 and CS4, lack of sister chromatid separation; CS5, defective chromosome segregation during anaphase; SA1, short spindles; SA2, spindles with a low MT density; SA3, poorly focused spindle poles; SA4 miscellaneous spindle defects; SC1, defective chromosome condensation and cytokinesis; SC2, multiple mitotic defects; SC1, defective in early cytokinesis; SC2, defective in late cytokinesis. The relative transcripts of the gene of the DM signature are also indicated according to their SEQ ID No.

								Contribution to the difference in score
Drosop hila gene symbol	Phenocluster	Human gene symbol	Human Gene name	Entrez Gene ID	Probeset	Transcri pt sequence ID No.	Pawitan z-score	Miller	Sotiriou-Desmedt	Wang
scra	CY2	ANLN	Anillin, actin binding protein	54443	222608_s_at	SEQ ID No. 1	4,39	2,35	-	-
dup	CS1	CDT1	Chromatin licensing and DNA replication factor 1	81620	228868_ x_at	SEQ ID No. 2	4,17	2,54	-	-
RnrS	CA	RRM2	Ribonucleotide reductase M2 polypeptide	6241	209773_s_at	SEQ ID No. 3	4,12	2,26	2,35	1,8
betaTub 56D	SA1	TUBB2C	Tubulin, beta 2C	10383	213726_x_at	SEQ ID No.4	4,06	0,69	0,54	0,34
betaTub 56D	SA1	TUBB2C	Tubulin, beta 2C	10383	208977_x_at	SEQ ID No.5	4,06	0,47	0,49	0,27
asp	SA3	ASPM	Asp (abnormal spindle) homolog, microcephaly associated (Drosophila)	259266	219918_s_at	SEQ ID No.6	3,99	2,56	1,66	1,93
CycB	CS3 & CS4	CCNB1	Cyclin B1	891	214710_s_at	SEQ ID No.7	3,95	2,43	2,32	1,18
pav	CY1	KIF23	Kinesin family member 23	9493	204709_s_at	SEQ ID No.8	3,91	2,51	1,23	0,98
RnrS	CA	RRM2	Ribonucleotide reductase M2 polypeptide	6241	201890_at	SEQ ID No.9	3,91	2,7	2,64	1,95
CG13298	CS3 & CS4	SF3B 14	Splicing factor 3B, 14 kDa subunit	51639	223416_at	SEQ ID No.10	3,85	0,38	-	-
gwl	CC3	MASTL	Microtubule associated serine/threonine kinase-like	84930	228468_at	SEQ ID No.11	3,73	0,69	-	-
tum	CY1	RACGAP 1	Rac GTPase activating protein 1	29127	222077_s_at	SEQ ID No.12	3,69	1,88	1,6	1,7
feo	CY1	PRC1	Protein regulator of cytokinesis 1	9055	218009_s_at	SEQ ID No.13	3,68	1,65	2,11	1,45
CycB	CS3 &CS4	CCNB1	Cyclin B1	891	228729_at	SEQ ID No.14	3,64	2,7	-	-
cdc2	SA4	CDC2	Cell division cycle 2, G1 to S and G2 to M	983	210559_s_at	SEQ ID No.15	3,57	1,76	2,08	0,77
cdc2	SA4	CDC2	Cell division cycle 2, G1 to S and G2 to M	983	203213_ at	SEQ ID No.16	3,55	2,79	1,9	1,36
fzy	CS5	CDC20	Cell division cycle 20 homolog (S. cerevisiae)	991	202870_s_at	SEQ ID No.17	3,52	2,14	2,04	1,38
Klp67A	SA4	KIF18A	Kinesin family member 18A	81930	221258_s_at	SEQ ID No.18	3,49	2,6	0,51	0,56
ncd	SA4	KIFC1	Kinesin family member C1	3833	209680_s_at	SEQ ID No.19	3,43	2,36	0,54	1,08
Top2	CC3	TOP2A	Topoisomerase (DNA) II alpha 170kDa	7153	201292_at	SEQ ID No.20	3,43	1,54	1,99	1,4
msps	SA1	CKAP5	Cytoskeleton associated protein 5	9793	212832_s_at	SEQ ID No.21	3,38	0,11	0,26	0,31
CG34438	CC2	NCAPG	Non-SMC condensin I complex, subunit G	64151	218662_s_at	SEQ ID No.22	3,37	1,54	1,79	0,72
U2A	CS3 & CS4	SNRPA1	Small nuclear ribonucleoprotein polypeptide A'	6627	216977_x_at	SEQ ID No.23	3,36	-0,09	0,82	0,43
CG34438	CC2	NCAPG	Non-SMC condensin I complex, subunit G	64151	218663_ at	SEQ ID No.24	3,36	2,03	1,57	1,06
Pros26.4	CS5	PSMC1	Proteasome (prosome, macropain) 26S subunit, ATPase, 1	5700	204219_s_at	SEQ ID No.25	3,35	0,36	0,37	0,18
CG3058	CS3 & CS4	TXNL4A	Thioredoxin-like 4A	10907	202836_s_at	SEQ ID No.26	3,27	0,91	0,52	0,12
polo	SC2	PLK1	Polo-like kinase 1 (Drosophila)	5347	202240 at	SEQ ID No.27	3,14	1,18	0,94	0,4
Mcm7	CC1	MCM7	Minichromosome maintenance complex component 7	4176	208795_s_at	SEQ ID No.28	3,13	0,51	0,6	0,38
Klp61F	SA4	KIF11	Kinesin family member 11	3832	204444_at	SEQ ID No.29	3,12	1,46	1,25	0,91
U2af38	CS3 & CS4	U2AF1	U2 small nuclear RNA auxiliary factor 1	7307	202858_at	SEQ ID No.30	3,08	0,94	0,33	0,23
cdc2	SA4	CDC2	Cell division cycle 2, G1 to S and G2 to M	983	203214_x_at	SEQ ID No.31	3,04	1,29	1,6	0,61
Ts	CA	TYMS	Thymidylate synthetase	7298	202589_ at	SEQ ID No.32	3,04	1,07	1,24	0,56
glu	CC2	SMC4	Structural maintenance of chromosomes 4	10051	201663_s_at	SEQ ID No.33	2,98	0,08	1,06	0,69
Mcm3	CC1	MCM3	Minichromosome maintenance complex component 3	4172	201555_at	SEQ ID No.34	2,90	0,45	0,33	-0,12
rod	CS2	KNTC1	Kinetochore associated 1	9735	206316_s_at	SEQ ID No.3 5	2,89	-0,29	0,49	0,7
dup	CS1	CDT1	Chromatin licensing and DNA replication factor 1	81620	209832_s_at	SEQ ID No.36	2,87	0,64	0,62	0,24
dnk	CA	TK2	Thymidine kinase 2,	7084	204227_s_at	SEQ ID	2,82	-0,13	-0,13	0,04
			mitochondrial			No.37
DDB1	CA	DDB1	Damage-specific DNA binding protein 1, 127kDa	1642	208619_ at	SEQ ID No.3 8	2,81	0,29	-0,11	0,17
CG6854	CA	CTPS	CTP synthase	1503	202613_ at	SEQ ID No.39	2,80	0,34	0,79	0,46
pbl	CY1	ECT2	Epithelial cell transforming sequence 2 oncogene	1894	234992_x_at	SEQ ID No.40	2,79	1,42	-	-
CG6937	SA3	MKI67IP	MKI67 (FHA domain) interacting nucleolar phosphoprotein	84365	224714_at	SEQ ID No.41	2,73	0,41	-	-
RfC40	CA	RFC2	Replication factor C (activator 1) 2, 40kDa	5982	203696_s_at	SEQ ID No.42	2,70	0,24	0,19	0,01
DNAprim	CA	PRIM2	Primase, DNA, polypeptide 2 (58kDa)	5558	205628_ at	SEQ ID No.43	2,68	1	0,02	0,14
pav	CY1	KIF23	Kinesin family member 23	9493	244427_at	SEQ ID No.44	2,49	0,31	-	-
SMC2	CC2	SMC2	Structural maintenance of chromosomes 2	10592	204240_s_at	SEQ ID No.45	2,46	0,22	0,92	0,28
CG7003	CA	MSH6	MutS homolog 6 (E. coli)	2956	202911_at	SEQ ID No.46	2,38	0,24	0,64	0,29
asp	SA3	ASPM	Asp (abnormal spindle) homolog, microcephaly associated (Drosophila)	259266	239002_at	SEQ ID No.47	2,37	1,35	-	-
RfC40	CA	RFC2	Replication factor C (activator 1) 2, 40kDa	5982	1053_at	SEQ ID No.48	2,33	0,11	0,22	-0,28
U2A	CS3 & CS4	SNRPA1	Small nuclear ribonucleoprotein polypeptide A'	6627	215722_s_at	SEQ ID No.49	2,33	0,13	0,57	0,12
CG4266	CS5	SFRS 15	Splicing factor, arginine/serine-rich 15	57466	226082_s_at	SEQ ID No.50	2,27	0,28	-	-
mus209	CA	PCNA	Proliferating cell nuclear antigen	5111	201202_at	SEQ ID	2,27	0,38	0,67	0,64
						No.51
Mcm7	CC1	MCM7	Minichromosome maintenance complex component 7	4176	210983_s_at	SEQ ID No.52	2,25	0,18	0,52	-0,07
asp	SA3	ASPM	Asp (abnormal spindle) homolog, microcephaly associated (Drosophila)	259266	232238_at	SEQ ID No.53	2,19	1,03	-	-
glu	CC2	SMC4	Structural maintenance of chromosomes	10051	201664_at	SEQ ID No.54	2,19	0,32	0,62	0,62
CG5931	CS3 & CS4	ASCC3L 1	Activating signal cointegrator 1 complex subunit 3-like 1	23020	200058_s_at	SEQ ID No.55	2,19	0,2	0,02	-0,09
DNAprim	CA	PRIM2	Primase, DNA, polypeptide 2 (58kDa)	5558	215708_s_at	SEQ ID No.56	2,16	0,06	0,11	-0,06
Bub3	CS2	BUB3	BUB3 budding uninhibited by benzimidazoles 3 homolog (yeast)	9184	201457_x_at	SEQ ID No.57	2,13	0,49	-0,02	-0,02
CG8241	CS3 & CS4	DHX8	DEAH (Asp-Glu-Ala-His) box polypeptide 8	1659	231184_at	SEQ ID No.58	1,94	-0,05	-	-
pbl	CY1	ECT2	Epithelial cell transforming sequence 2 oncogene	1894	219787_s_at	SEQ ID No.59	1,94	0,73	0,55	0,48
CG6876	CS3 &CS4	PRPF31	PRP31 pre-mRNA processing factor 31 homolog (S. cerevisiae)	26121	202407_s_at	SEQ ID No.60	1,90	0,33	0,37	-0,4
CG7003	CA	MSH6	MutS homolog 6 (E. coli)	2956	211450_s_at	SEQ ID No.61	1,88	0,29	0,61	-0,12
Top2	CC3	TOP2A	Topoisomerase (DNA) II alpha 170kDa	7153	201291_s_at	SEQ ID No.62	1,78	1,1	1,3	0,81
CG4266	CS5	SFRS 15	Splicing factor, arginine/serine-rich 15	57466	233753_at	SEQ ID No.63	1,76	0,33	-	-
RpA-70	CA	RPA1	Replication protein A1, 70kDa	6117	201529_s_at	SEQ ID No.64	1,73	-0,24	-0,06	-0,21
CG2685	CA	WBP11	WW domain binding protein 11	51729	217821_s_at	SEQ ID	1,72	-0,43	0,05	-0,24
						No.65
1(2)NC1 36	CA	CNOT3	CCR4-NOT transcription complex, subunit 3	4849	211141_s_at	SEQ ID No.66	1,68	0,5	-0,03	0,01
CG2685	CA	WBP11	WW domain binding protein 11	51729	217822_ at	SEQ ID No.67	1,64	-0,03	0,04	0,14
Taf6	CA	TAF6	TAF6 RNA polymerase II, TATA box binding protein (TBP)-associated factor, 80kDa	6878	203572_s_at	SEQ ID No.68	1,62	-0,06	-0,05	0,06
Nipped-A	SA2	TRRAP	Transformation/transcription domain-associated protein	8295	202642_s_at	SEQ ID No.69	1,61	0,02	0,01	0,28
Orc5	CA	ORC5L	Origin recognition complex, subunit 5-like (yeast)	5001	211212_s_at	SEQ ID No.70	1,58	-0,02	0,08	0,07
U2A	CS3 & CS4	SNRPA1	Small nuclear ribonucleoprotein polypeptide A'	6627	206055_s_at	SEQ ID No.71	1,56	0,24	0,24	0,3
CG1859 1	CS3 & CS4	SNRPE	Small nuclear ribonucleoprotein polypeptide E	6635	203316_s_at	SEQ ID No.72	1,54	0,28	0,05	0,16
Nipped-A	SA2	TRRAP	Transformation/transcription domain-associated protein	8295	214908_s_at	SEQ ID No.73	1,52	-0,14	-0,05	-0,14
CG8950	SA1	GTF3C3	General transcription factor IIIC, polypeptide 3, 102kDa	9330	218343_s_at	SEQ ID No.74	1,50	0,07	0,15	0,06
okr	CA	RAD54L	RAD54-like (S. cerevisiae)	8438	204558_at	SEQ ID No.75	1,49	0,05	0,26	0,64
Grip75	SA2	TUBGCP 4	Tubulin, gamma complex associated protein 4	27229	211337_s_at	SEQ ID No.76	1,47	-0,1	0,22	0,01
CG3605	CS3 &CS4	SF3B2	Splicing factor 3b, subunit 2, 145kDa	10992	200619_at	SEQ ID No.77	1,44	0,07	0,05	0,07
sti	CY2	CIT	Citron (rho-interacting, serine/threonine kinase 21)	11113	212801_at	SEQ ID No.78	1,43	0,01	0,02	0,07
Orc5	CA	ORC5L	Origin recognition complex, subunit 5-like (yeast)	5001	204957_at	SEQ ID No.79	1,36	0,05	0,12	0,1
CG4266	CS5	SFRS 15	Splicing factor, arginine/serine-rich 15	57466	222311 _s_at	SEQ ID No.80	1,36	-0,1	0,12	0,02
CG1035 4	CA	XRN2	5'-3' exoribonuclease 2	22803	233878_s_at	SEQ ID No.81	1,30	-0,08	-	-
U2af3 8	CS3 & CS4	U2AF1	U2 small nuclearRNA auxiliary factor 1	7307	242499_at	SEQ ID No.82	1,27	0,11	-	-
pbl	CY1	ECT2	Epithelial cell transforming sequence 2 oncogene	1894	237241_at	SEQ ID No.83	1,23	0,05	-	-
CG1035 4	CA	XRN2	5'-3' exoribonuclease 2	22803	223002_s_at	SEQ ID No.84	1,21	-0,07	-	-
ida	CS5	ANAPC5	Anaphase promoting complex subunit 5	51433	208721_s_at	SEQ ID No.85	1,14	0,03	0,09	-0,15
Dp	CA	TFDP2	Transcription factor Dp-2 (E2F dimerization partner 2)	7029	203588_s_at	SEQ ID No.86	1,14	0,16	0,24	-0,13
Sas-4	SA3	CENPJ	Centromere protein J	55835	220885_s_at	SEQ ID No.87	1,12	0,03	0,05	-0,05
DNApri m	CA	PRIM2	Primase, DNA, polypeptide 2 (58kDa)	5558	215709_at	SEQ ID No.88	1,10	0,02	-0,01	0,04
CG6937	SA3	MKI67IP	MKI67 (FHA domain) interacting nucleolar phosphoprotein	84365	224713_at	SEQ ID No.89	1,04	0,06	-	-
ial	SC1	AURKC	Aurora kinase C	6795	211107_s_at	SEQ ID No.90	1,02	0,1	-0,02	-0,08
CG6876	CS3 & CS4	PRPF31	PRP31 pre-mRNA processing factor 31 homolog (S. cerevisiae)	26121	202408_s_at	SEQ ID No.91	1,01	0,29	0,1	0,08
Sas-4	SA3	CENPJ	Centromere protein J	55835	223513_at	SEQ ID No.92	0,94	0,03	-	-
CG2260	CA	WDR46	WD repeat domain 46	9277	209196_at	SEQ ID No.93	0,91	0,21	0,01	-0,15
U2af50	CS3 & CS4	U2AF2	U2 small nuclear RNA auxiliary factor 2	11338	218382_s_at	SEQ ID No.94	0,82	0,07	0,02	-0,21
ida	CS5	ANAPC5	Anaphase promoting complex subunit 5	51433	200098_s_at	SEQ ID No.95	0,82	0,01 1	0,02	0,09
Top2	CC3	TOP2A	Topoisomerase (DNA) II alpha 170kDa	7153	237469_at	SEQ ID No.96	0,79	0,24	-	-
ida	CS5	ANAPC5	Anaphase promoting complex subunit 5	51433	208722_s_at	SEQ ID No.97	0,77	-0,1	-0,01	0,06
CG1694 1	CS3 & CS4	SF3A1	Splicing factor 3a, subunit 1, 120kDa	10291	201357_s_at	SEQ ID No.98	0,71	-0,09	-0,01	-0,11
Mtor	CA	TPR	Translocated promoter region (to activated MET oncogene)	7175	215220_s_at	SEQ ID No.99	0,68	-0,16	-0,02	-0,12
CG6015	CS3 &CS4	CDC40	Cell division cycle 40 homolog (S. cerevisiae)	51362	203377_s_at	SEQ ID No.100	0,65	0,02	0	0,1
eIF-3p66	SA1	EIF3D	Eukaryotic translation initiation factor 3, subunit D	8664	200005_at	SEQ ID No.101	0,63	0,05	-0,04	-0,04
U2af3 8	CS3 & CS4	U2AF1	U2 small nuclear RNA auxiliary factor 1	7307	232141_at	SEQ ID No.102	0,58	0,03	-	-
RpA-70	CA	RPA1	Replication protein A1, 70kDa	6117	201528_at	SEQ ID No.103	0,57	-0,03	0,01	0,07
Orc2	CC3	ORC2L	Origin recognition complex, subunit 2-like (yeast)	4999	204853_at	SEQ ID No.104	0,56	0,05	0,01	0,03
Nnp-1	SA1	RRP1B	Ribosomal RNA processing 1 homolog B (S. cerevisiae)	23076	212844_at	SEQ ID No.105	0,54	0,05	0,04	0,01
CAP-D2	CC2	NCAPD2	Non-SMC condensin I complex, subunit D2	9918	201774_s_at	SEQ ID No.106	0,52	0,03	0,12	-0,03
Sas-4	SA3	CENPJ	Centromere protein J	55835	234023_s_at	SEQ ID No.107	0,44	0,06	-	-
Mtor	CA	TPR	Translocated promoter region (to activated MET oncogene)	7175	201731_s_at	SEQ ID No.108	0,44	0	-0,07	0,01
Orc5	CA	ORC5L	Origin recognition complex, subunit 5-like (yeast)	5001	211213_at	SEQ ID No.109	0,44	0,09	0	-0,01
tho2	SA1	THOC2	THO complex 2	57187	226628_at	SEQ ID No.110	0,40	0	-	-
kin17	CS5	KIN	KIN, antigenic determinant of recA protein homolog (mouse)	22944	205664_at	SEQ ID No.111	0,34	0,03	0,04	0,01
ida	CS5	ANAPC5	Anaphase promoting complex subunit 5	51433	211036_x_at	SEQ ID No.112	0,33	-0,02	0	0,06
cul-4	CA	CUL4B	Cullin 4B	8450	210257_x_at	SEQ ID No.113	0,29	-0,03	0,03	0
Trip1	SA1	EIF3I	Eukaryotic translation initiation factor 3, subunit I	8668	208756_at	SEQ ID No.114	0,26	0,01	-0,01	-0,04
SMC1	CA	SMC1A	Structural maintenance of chromosomes 1A	8243	201589_at	SEQ ID No.115	0,26	0,03	0,05	0,05
Eb1	SA1	MAPRE3	Microtubule-associated protein, RP/EB family, member 3	22924	203842_s_at	SEQ ID No.116	0,25	-0,01	0,01	-0,02
ida	CS5	ANAPC5	Anaphase promoting complex subunit 5	51433	239651_at	SEQ ID No.117	0,25	-0,01	-	-
Dp	CA	TFDP2	Transcription factor Dp-2 (E2F dimerization partner 2)	7029	203589_s_at	SEQ ID No.118	0,25	-0,03	0,01	-0,01
CG1694 1	CS3 & CS4	SF3A1	Splicing factor 3a, subunit 1, 120kDa	10291	227516_at	SEQ ID No.119	0,19	-0,06	-	-
CG8241	CS3 &CS4	DHX8	DEAH (Asp-Glu-Ala-His) box polypeptide 8	1659	227079 at	SEQ ID No.120	0,15	0	-	-
CG8241	CS3 & CS4	DHX8	DEAH (Asp-Glu-Ala-His) box polypeptide 8	1659	203334_at	SEQ ID No.121	0,08	-0,03	0	0
sti	CY2	CIT	Citron (rho-interacting, serine/threonine kinase 21)	11113	242872_at	SEQ ID No.122	0,07	0,02	-	-
Nnp-1	SA1	RRP1B	Ribosomal RNA processing 1 homolog B (S. cerevisiae)	23076	212846_at	SEQ ID No.123	0,06	0,02	0,01	0,01
CG6686	CA	SART1	Squamous cell carcinoma antigen recognized by T cells	9092	200051_at	SEQ ID No.124	0,02	0	0
Ts	CA	TYMS	Thymidylate synthetase	7298	217684_at	SEQ ID No. 125	0,01	0	0	0
CG1939	CA	DCAKD	Dephospho-CoA kinase domain containing	79877	221225_at	SEQ ID No.126	0,01	0	2,29E-05	0
SMC2	CC2	SMC2	Structural maintenance of chromosomes 2	10592	213253_ at	SEQ ID No. 127	-0,03	-0,01	-0,01	0
Eb1	SA1	MAPRE3	Microtubule-associated protein, RP/EB family, member 3	22924	214270_s_at	SEQ ID No. 128	-0,05	0,01	0	0
CG1141 9	CS5	ANAPC1 0	Anaphase promoting complex subunit 10	10393	207845_s_at	SEQ ID No. 129	-0,07	0	-0,01	-0,01
Eb1	SA1	MAPRE3	Microtubule-associated protein, RP/EB family, member 3	22924	203841_x_at	SEQ ID No.130	-0,08	0,01	0	0
U2af38 8	CS3 & CS4	U2AF1	U2 small nuclear RNA auxiliary factor 1	7307	231904_at	SEQ ID No. 131	-0,10	-0,01	-	-
tho2	SA1	THOC2	THO complex 2	57187	226626_ at	SEQ ID No.132	-0,11	0	-	-
Eb1	SA1	MAPRE3	Microtubule-associated protein, RP/EB family, member 3	22924	229682_at	SEQ ID No.133	-0,13	-0,03	-	-
CG3058	CS3 & CS4	TXNL4A	Thioredoxin-like 4A	10907	202835_ at	SEQ ID No.134	-0,13	-0,02	-0,01	-0,01
CG5931	CS3 & CS4	ASCC3L 1	Activating signal cointegrator 1 complex subunit 3-like 1	23020	232931_at	SEQ ID No.135	-0,15	-0,02	-	-
CG1859 1	CS3 & CS4	SNRPE	Small nuclear ribonucleoprotein polypeptide E	6635	231112_at	SEQ ID No.136	-0,15	0,03	-	-
CG1041 8	CS3 & CS4	LSM2	LSM2 homolog, U6 small nuclear RNA associated (S. cerevisiae)	57819	209449_at	SEQ ID No.137	-0,19	-0,01	-0,01	0,01
1(2)NC1 36	CA	CNOT3	CCR4-NOT transcription complex, subunit 3	4849	203239_s_at	SEQ ID No.138	-0,20	-0,03	0,01	0,01
Bub3	CS2	BUB3	BUB3 budding uninhibited by benzimidazoles 3 homolog (yeast)	9184	209974_s_at	SEQ ID No.139	-0,21	-0,05	0,01	-0,03
CG1939	CA	DCAKD	Dephospho-CoA kinase domain containing	79877	221224_s_at	SEQ ID No.140	-0,24	0,06	-0,01	0,09
glu	CC2	SMC4	Structural maintenance of chromosomes 4	10051	215623_x_at	SEQ ID No.141	-0,24	-0,02	-0,03	-0,05
CG4266	CS5	SFRS 15	Splicing factor, arginine/serine-rich 15	57466	243759_at	SEQ ID No.142	-0,28	-0,02	-	-
mit(1)15	CS2	ZW10	ZW10, kinetochore associated, homolog (Drosophila)	9183	204812_at	SEQ ID No.143	-0,32	0,01	0,01	0
kin17	CS5	KIN	KIN, antigenic determinant of recA protein homolog (mouse)	22944	236887_at	SEQ ID No.144	-0,34	0,03	-	-
CG4785	CA	C15orf44	Chromosome 15 open reading frame 44	81556	221265_s_at	SEQ ID No.145	-0,34	-0,02	0	0
U2af50	CS3 & CS4	U2AF2	U2 small nuclear RNA auxiliary factor 2	11338	229508_at	SEQ ID No.146	-0,35	-0,08	-	-
DNApol alpha18 0	CA	POLA1	Polymerase (DNA directed), alpha 1, catalytic subunit	5422	204835_at	SEQ ID No.147	-0,37	0,05	-0,03	-0,01
Bub3	CS2	BUB3	BUB3 budding uninhibited by benzimidazoles 3 homolog (yeast)	9184	229827_at	SEQ ID No.148	-0,37	-0,09	-	-
CG1420	CS3 & CS4	SLU7	SLU7 splicing factor homolog (S. cerevisiae)	10569	231718_at	SEQ ID No.149	-0,38	-0,02	-	-
CG5931	CS3 & CS4	ASCC3L 1	Activating signal cointegrator 1 complex subunit 3-like 1	23020	214982_at	SEQ ID No.150	-0,38	0,1	-0,01	-0,02
CG1420	CS3 & CS4	SLU7	SLU7 splicing factor homolog (S. cerevisiae)	10569	227990_at	SEQ ID No.151	-0,41	0,03	-	-
cul-4	CA	CUL4B	Cullin 4B	8450	202213_s_at	SEQ ID No.152	-0,43	0,03	-0,02	0,01
CG7003	CA	MSH6	MutS homolog 6 (E. coli)	2956	240148_at	SEQ ID No.153	-0,45	-0,05	-	-
Int6	SA1	EIF3E	Eukaryotic translation initiation factor 3, subunit E	3646	208697_s_at	SEQ ID No.154	-0,45	-0,03	-0,03	-0,03
U2af50	CS3 & CS4	U2AF2	U2 small nuclear RNA auxiliary factor 2	11338	214171_s_at	SEQ ID No.155	-0,48	0,05	-0,01	0,02
dnk	CA	TK2	Thymidine kinase 2, mitochondrial	7084	204277_s_at	SEQ ID No.156	-0,49	0,11	0,01	0,07
gamma Tub23C	SA2	TUBG2	Tubulin, gamma 2	27175	203894_at	SEQ ID No.157	-0,53	0,02	0	0,01
CG1205 0	CA	WDR75	WD repeat domain 75	84128	224721_at	SEQ ID No.158	-0,54	-0,02	-	-
C12.1	CA	CWC15	CWC15 homolog (S. cerevisiae)	51503	223067_at	SEQ ID No.159	-0,55	-0,03	-	-
CG8233	CS3 & CS4	KIAA131 0	KIAA1310	55683	224318_s_at	SEQ ID No.160	-0,56	0,05	-	-
U2af50	CS3 & CS4	U2AF2	U2 small nuclear RNA auxiliary factor 2	11338	218381_s_at	SEQ ID No.161	-0,56	-0,14	0	-0,08
CG1694 1	CS3 & CS4	SF3A1	Splicing factor 3a, subunit 1, 120kDa	10291	216457_s_at	SEQ ID No.162	-0,56	0,05	0	-0,01
CG8950	SA1	GTF3C3	General transcription factor IIIC, polypeptide 3, 102kDa	9330	222604_at	SEQ ID No.163	-0,57	0,01	-	-
CG1234	SA1	NOC3L	Nucleolar complex associated 3 homolog (S. cerevisiae)	64318	218889_at	SEQ ID No.164	-0,57	-0,04	-0,02	0,03
cul-4	CA	CUL4B	Cullin 4B	8450	215997_s_at	SEQ ID No.165	-0,63	0	-0,01	0
Caf1	SC1	RBBP7	Retinoblastoma binding protein 7	5931	201092_at	SEQ ID No.166	-0,64	-0,08	-0,07	-0,06
1(2)NC1 36	CA	CNOT3	CCR4-NOT transcription complex, subunit 3	4849	229143_at	SEQ ID No.167	-0,71	-0,07	-	-
NiPp1	SA3	PPP1R8	Protein phosphatase 1, regulatory (inhibitor) subunit 8	5511	207830_s_at	SEQ ID No.168	-0,71	0,02	0	0
CG1075 4	CS3 & CS4	SF3A2	Splicing factor 3a, subunit 2, 66kDa	8175	209381_x_at	SEQ ID No.169	-0,73	-0,14	0,06	0,07
CG7757	CA	PRPF3	PRP3 pre-mRNA processing factor 3 homolog (S. cerevisiae)	9129	202251_at	SEQ ID No.170	-0,74	0,13	-0,04	-0,11
dnk	CA	TK2	Thymidine kinase 2, mitochondrial	7084	240300_at	SEQ ID No.171	-0,76	-0,04	-	-
dnk	CA	TK2	Thymidine kinase 2, mitochondrial	7084	204276_at	SEQ ID No.172	-0,76	0,17	0,02	0,05
CG1694 1	CS3 & CS4	SF3A1	Splicing factor 3a, subunit 1, 120kDa	10291	201356_at	SEQ ID No.173	-0,77	0,11	-0,01	0,04
Bub3	CS2	BUB3	BUB3 budding uninhibited by benzimidazoles 3 homolog (yeast)	9184	201458_s_at	SEQ ID No.174	-0,83	-0,06	0,04	-0,08
eIF3-S10	SA1	EIF3A	Eukaryotic translation initiation factor 3, subunit A	8661	200595_s_at	SEQ ID No.175	-0,84	0	0,04	0
CG1075 4	CS3 &CS4	SF3A2	Splicing factor 3a, subunit 2, 66kDa	8175	37462_i_at at	SEQ ID No.176	-0,84	-0,09	0,1	0,02
CG6015	CS3 &CS4	CDC40	Cell division cycle 40 homolog (S. cerevisiae)	51362	203376_at	SEQ ID No.177	-0,93	0	0,09	-0,09
Bub3	CS2	BUB3	BUB3 budding uninhibited by benzimidazoles 3 homolog (yeast)	9184	201456_s_at	SEQ ID No.178	-0,96	-0,03	0,1	-0,08
SMC1	CA	SMC1A	Structural maintenance of chromosomes 1A	8243	239688_at	SEQ ID No.179	-1,01	-0,21	-	-
CG6197	CA	XAB2	XPA binding protein 2	56949	218110_at	SEQ ID No.180	-1,05	-0,1	0,12	0,11
U2A	CS3 & CS4	SNRPA1	Small nuclear ribonucleoprotein polypeptide A'	6627	242146_at	SEQ ID No.181	-1,11	0,06	-	-
Mtor	CA	TPR	Translocated promoter region (to activated MET oncogene)	7175	228709_at	SEQ ID No.182	-1,16	-0,03	-	-
CG8233	CS3 & CS4	KIAA131 0	KIAA1310	55683	220950_s_at	SEQ ID No.183	-1,18	-0,18	0,09	0,09
CG6876	CS3 & CS4	PRPF31	PRP31 pre-mRNA processing factor 31 homolog (S. cerevisiae)	26121	214380_at	SEQ ID No.184	-1,19	0,17	-0,04	0,01
Cap	CC1	SMC3	Structural maintenance of chromosomes	9126	209259_s_at	SEQ ID No.185	-1,26	-0,04	-0,09	-0,09
eIF3-S10	SA1	EIF3A	Eukaryotic translation initiation factor 3, subunit A	8661	200597_at	SEQ ID No.186	-1,36	0,06	0,14	-0,03
CG7003	CA	MSH6	MutS homolog 6 (E. coli)	2956	211449_at	SEQ ID No.187	-1,38	-0,2	0,01	0,02
CG8233	CS3 & CS4	KIAA131 0	KIAA1310	55683	223756_at	SEQ ID No.188	-1,39	-0,27	-	-
Dcp-1	CA	CASP7	Caspase 7, apoptosis-related cysteine peptidase	840	207181_s_at	SEQ ID No.189	-1,47	-0,11	0,15	0,14
CG1939	CA	DCAKD	Dephospho-CoA kinase domain containing	79877	224522_s_at	SEQ ID No.190	-1,54	0,17	-	-
CG1729 3	SA1	WDR82	WD repeat domain 82	80335	201934_at	SEQ ID No.191	-1,55	-0,1	0,24	0,02
woc	CA	ZMYM4	Zinc finger, MYM-type 4	9202	202049_s_at	SEQ ID No.192	-1,58	0,01	0,1	0,1
Cap	CC1	SMC3	Structural maintenance of chromosomes 3	9126	209257_s_at	SEQ ID No.193	-1,62	0,18	-0,1	0,04
CG2807	CS3 & CS4	SF3B1 1	Splicing factor 3b, subunit 1, 155kDa	23451	201071_x_at	SEQ ID No.194	-1,63	0,16	0,03	-0,02
CG6480	CA	FRG1	FSHD region gene 1	2483	204145_at	SEQ ID No.195	-1,69	-0,19	-0,13	-0,01
WOC	CA	ZMYM4	Zinc finger, MYM-type 4	9202	202050_s_at	SEQ ID No.196	-1,69	0,14	0,15	-0,03
CG4266	CS5	SFRS 15	Splicing factor, arginine/serine-rich 15	57466	222310_at	SEQ ID No.197	-1,71	-0,14	-0,09	-0,15
ik2	SA1	TBK1	TANK-binding kinase 1	29110	218520_at	SEQ ID No.198	-1,71	0,15	-0,22	-0,1
tho2	SA1	THOC2	THO complex 2	57187	212994_at	SEQ ID No.199	-1,75	-0,01	-0,1	0,29
CG6937	SA3	MKI67IP	MKI67 (FHA domain) interacting nucleolar phosphoprotein	84365	234167_at	SEQ ID No.200	-1,76	0,44	-	-
noi	CA	SF3A3	Splicing factor 3a, subunit 3, 60kDa	10946	203818_s_at	SEQ ID No.201	-1,92	0,02	-0,03	0,17
eIF3-S10	SA1	EIF3A	Eukaryotic translation initiation factor 3, subunit A	8661	200596_s_at	SEQ ID No.202	-1,95	-0,09	0,09	0,34
CG2807	CS3 & CS4	SF3B1	Splicing factor 3b, subunit 1, 155kDa	23451	214305_s_at	SEQ ID No.203	-2,14	0,13	0,1	0,11
Mtor	CA	TPR	Translocated promoter region (to activated MET oncogene)	7175	201730_s_at	SEQ ID No.204	-2,29	-0,01	0,13	-0,02
tho2	SA1	THOC2	THO complex 2	57187	222122_s_at	SEQ ID No.205	-2,29	0,06	-0,23	-0,03
cdc2	SA4	CDC2	Cell division cycle 2, G1 to S and G2 to M	983	231534_at	SEQ ID No.206	-2,31	-0,02	-	-
SMC1	CA	SMC1A	Structural maintenance of chromosomes 1A	8243	217555_at	SEQ ID No.207	-2,31	-0,11	-0,05	0,06
CG2807	CS3 & CS4	SF3B1	Splicing factor 3b, subunit 1, 155kDa	23451	211185_s_at	SEQ ID No.208	-2,57	0,08	0,06	-0,03
Cap	CC1	SMC3	Structural maintenance of chromosomes 3	9126	209258_s_at	SEQ ID No.209	-2,62	-0,09	-0,2	0,26
Dp	CA	TFDP2	Transcription factor Dp-2 (E2F dimerization partner 2)	7029	226157_at	SEQ ID No.210	-2,67	0,5	-	-
Su(var)2 -10	CA	PIAS1	Protein inhibitor of activated STAT, 1	8554	217864_s_at	SEQ ID No.211	-2,71	0,13	0,05	0,36
cul-4	CA	CUL4B	Cullin 4B	8450	202214_s_at	SEQ ID No.212	-2,96	0,29	-0,17	0,27
woc	CA	ZMYM4	Zinc finger, MYM-type 4	9202	202051_s_at	SEQ ID No.213	-3,10	-0,04	0,22	-0,09
CG2807	CS3 &CS4	SF3B1	Splicing factor 3b, subunit 1, 155kDa	23451	201070_x_at	SEQ ID No.214	-3,46	0,27	0,04	0,22
Grip75	SA2	TUBGCP 4	Tubulin, gamma complex associated protein 4	27229	213266_at	SEQ ID No.215	-3,56	0,68	0,52	-0,01
Su(var)2 -10	CA	PIAS1	Protein inhibitor of activated STAT, 1	8554	217862_at	SEQ ID No.216	-3,84	0,45	0,09	0,35
Su(var)2 -10	CA	PIAS1	Protein inhibitor of activated STAT, 1	8554	217863_at	SEQ ID No.217	-4,30	0,2	0	-0,15

Collectively, the genes in Table I constitute the DM signature. The remaining 34 Drosophila genes identified in the screen [16] were not included in the DM signature because they did not have an unambiguous human homologue in Homologene (Release 62). The DM signature shares very few genes with pre-existing signatures. We considered the top-down 70-gene signature [3] and several bottom-up signatures based on various aspects of cancer biology: the Wound signature [5,6]: the ES signature [12]; the IGS signature [14] the Hypoxia signatures of Sung et al. [8] and Winter et al. [9]; the Proliferation signature of Starmans et al. [11]; the proliferation/immune response/RNA splicing (Module) signature [13] and the chromosomal instability (CIN) signature [15]. The number of genes that the DM signature shares with the 70-gene, ES, IGS, Wound and Hypoxia signatures is extremely small. The overlap is higher with the Module, Proliferation and CIN signatures, but none of these signatures shares more that 20% of its genes with the DM signature (Table III).

Table III. The DM signature shares very few genes with other major cancer signatures

*Signature*	# of genes in the signature	*Genes in common with the DM signature*
Module	261	18 (6.9%)
CIN	71	14 (19.7)
ES	1029	14 (1.4%)
Wound	371	6 (1.6%)
Proliferation	52	6 (11.5%)
70-gene	61	2 (3.3%)
Hypoxia (Winter)	92	2 (2.2%)
IGS	175	2 (1,1%)
Hypoxia (Sung)	126	1 (0.8%)

Of the 108 human genes, 25 are included in the list of genes periodically expressed during the cell cycle in HeLa cells {pmid: 12058064}, compared to 5.8 expected by chance (P=2.2E-10): therefore, as expected, the human orthologs of genes that display a mitotic phenotype in the fly tend to be regulated by the cell cycle also in human.
For each dataset and each signature the same analysis as the one shown in Fig. 1 was performed and the value of P log-rank was compared to that calculated for the DM signature. In agreement with previous studies, the vast majority of the signatures show a good predictive value in the majority of the datasets (Fig. 3). The signature DM has a higher performance (in terms of P-value in the log-rank test) when compared to all other signatures in the majority of datasets (Fig. 3). Further, the DM signature has a statistically significant predictive power in all datasets and the lower overall P-value.

The prognostic value of the DM signature

For assessment of the predictive power and robustness of the DM signature the authors used six publicly available breast cancer datasets: (i) NKI, which contains expression data from primary breast tumors for 295 consecutive, relatively young (age < 52 yrs) patients [4]; (ii) Pawitan, which includes data from 159 consecutive breast cancer patients [20]; (iii) Miller, with data from 251 patients selected from a consecutive series based on the quality of the material [21]; (iv) Desmedt and (v) Wang, which contains expression data from 198 and 286 lymph-node negative, systemically untreated patients, respectively [22,23]; (vi) Sotiriou, which includes 189 invasive breast carcinomas [24]. Due to the presence of common samples, the authors merged the Desmedt and Sotiriou datasets into a single one and removed from it the patients that were also included in the Miller dataset. All datasets contain both ER-positive and ER-negative samples.
Although most of these gene expression data were generated using the same microarray platform, and could in principle be merged in a single dataset as recently described [13], the authors evaluated the DM signature on the individual datasets. The authors chose this approach because the robustness of a gene signature on independent datasets is an important criterion for validation of its predictive power. In the authors' prognostic power analysis, they used relapse-free survival times when available, or overall survival times otherwise. Because three genes of the DM signature (H3F3A, PPAN-P2RY11 and KIF4) were not represented in the Affymetrix platform, the authors performed their analyses on 105 genes. For each dataset, patients were divided into two groups based on the expression profiles of the genes in the DM signature using hierarchical clustering. Differences in survival probability between the two groups were then evaluated with a standard log-rank test on Kaplan-Meier curves. Fig. 1 shows that the differences in survival are statistically significant for all datasets considered.
As mentioned above, the DM signature contains two broad classes of genes, namely 72 mitotic genes (71 in platform) and 36 genes required for the maintenance of chromosome integrity (34 in platform). To determine the relative contribution of these two gene classes to the predictive power of the DM signature, the authors performed the analysis using the two categories of genes separately. Both gene groups turned out to be independently predictive of survival (Fig. 2). However the predictive power of the global signature was higher in all cases.

The authors also asked whether the DM signature is predictive of survival in other tumors besides breast cancer. Using the hierarchical clustering approach described above, the authors found that the DM signature is predictive of survival in a large lung cancer dataset [18] (P=3e-6, Fig. 6) and in a glioma dataset [19] (P=0.0170, Fig. 7). However, the DM signature is not significantly predictive in other lung cancer [27] and glioma [28] datasets, or in renal [29] and ovarian [27] cancer datasets. The p-values of the log-rank tests for non-breast datasets are reported in Table IV.

Table IV - Predictive power of the DM signature in cancers other than breast. The p-values obtained from the log-rank test when comparing the cumulative probability of survival of clusters of patients in other types of cancer.

dataset	Log (p-value)
Glioma (Freij e)	1,77 **	* P > 0.05
Glioma (Phillips)	0,27 *	** 0.05 > P > 0.01
Lung (Bild)	0,21 *	***P<0.01
Lung (Shedden)	3,52 ***
Ovarian (Bild)	0,57 *
Renal (Zhao)	1,12 *

Evaluation of a prognostic score for the DM signature

Subdivision of patients into risk groups using the unsupervised clustering-based approach described above allows assessment of the predictive power of a gene signature, but does not allow specificity (fraction of low-risk patients correctly classified) and sensitivity (fraction of high-risk patients correctly classified) to be tuned according to specific requirements. However, such tuning is important in clinical applications, because the misclassification of a high-risk patient is potentially more harmful than the misclassification of a low-risk patient. Indeed, the 70-gene signature [3], which is used in clinical practice, assigns a risk score to each patient; patients are then classified based on a score threshold that can be tuned to obtain the desired compromise between specificity and sensitivity. Scalable prognostic scores, each computed from gene expression data with a specific algorithm, have been previously defined also for the Wound [6], IGS [14], Proliferation [11], CIN [15] and Hypoxia [9] signatures.
The authors determined a scalable prognostic score for the DM signature, using a procedure similar to that employed by Wang and co-workers [22]. The authors define the DM prognostic score as the sum of the logarithmic expression values of the signature genes, each multiplied by its z-score. The Cox z-score measures the correlation between the expression pattern of a gene and survival of the patient. A positive (negative) z-score indicates negative (positive) correlation between the gene expression level and patient's survival time.
The authors used the Pawitan dataset as training set and computed the Cox z-scores for the Affymetrix probesets associated to the DM signature (the z-scores of all probesets are shown in Table II). The distribution of these z-scores is consistently shifted towards positive values compared to the distribution of the z-scores of all genes represented on the microarrays (P-values between 1.1e-6 and 3.3e-15 from one-sided Mann-Whitney U test) (Fig. 4). Thus, as expected for proliferation-related genes, for most genes in the DM signature an increased expression level is negatively correlated with survival.
The authors then compared the DM signature score with the scores of 6 other scalable signatures for performance in predicting cancer outcome at 5 years. For this analysis the authors used ROC curves generated with the Affymetrix datasets not employed for training (Miller, Sotiriou-Desmedt and Wang). The scores of the CIN [15], Proliferation [11], 70-gene [3], Wound [6], IGS [14], and Hypoxia [9] signatures were computed as described in the respective references, after mapping the genes to the Affymetrix platform (see Methods for details). As shown in Fig. 5, the predictive power of the 3 proliferation-based signatures (DM, CIN and Proliferation), measured by the Area Under ROC Curves (AUC), is very similar in all datasets and systematically higher than that of the 70-gene, Wound, IGS, or Hypoxia signature.

Since the DM signature and the two other proliferation-based signatures perform similarly in predicting outcome at 5 years, as shown by the AUC values in Fig. 5, the authors compared their performance in greater detail at three values of the sensitivity (percentage of poor-outcome patients that are classified correctly by the signature). The results are shown in Tab. V.

Table V. Comparison of the performance of the proliferation-based signatures

90% sensitivity	DM		CIN		Proliferation
	P value	Spec.	P value	Spec.	P value	Spec.
Miller	2.26E-04	0.318	5.44E-04	0.352	4.89E-04	0.352
Sotiriou-Desmedt	4.44E-03	0.335	0.0312	0.329	0.0124	0.329
Wang	4.08E-03	0.226	0.0114	0.260	0.015	0.227

70% sensitivity	DM		CIN		Proliferation
	P value	Spec.	P value	Spec.	P value	Spec.
Miller	1.77E-04	0.614	7.63E-03	0.523	3.02E-03	0.562
Sotiriou-Desmedt	4.51E-04	0.613	4.25E-04	0.600	1.24E-03	0.574
Wang	4.25E-04	0.547	5.58E-04	0.547	1.19E-03	0.536

50% sensitivity	DM		CIN		Proliferation
	P value	Spec.	P value	Spec.	P value	Spec.
Miller	3.91E-04	0.733	8.81E-04	0.705	1.42E-03	0.716
Sotiriou-Desmedt	0.138	0.697	0.134	0.722	0.161	0.690
Wang	6.85E-03	0.669	2.41E-03	0.691	0.022	0.641

Tab. V reports for each signature and each dataset the specificity (percentage of correct classifications among patients classified as poor-outcome), and the P-value of the log-rank test between the two groups of patients. These two parameters have different interpretations: while the specificity refers to the ability of the signature to predict the outcome specifically at the 5-years endpoint, the P-value takes into account the complete survival data, and thus measures the ability to stratify the patients over the whole time range.

The results show that the DM and CIN signatures tend to perform better than the Proliferation one at all tested sensitivity values. DM performs slightly better than CIN at higher sensitivities, especially in terms of P-value. These differences in performance between the three signatures are driven by percentages of discordantly classified patients ranging from ∼2% to ∼10%. The number of discordantly classified patients in the three datasets is reported in Table VI.

Table VI: Cox multivariate analysis for various breast cancer datasets. The DM score is a predictor of survival independent of clinical and histological parameters commonly used in patient stratification. The table shows the odd-ratio and P-value obtained from a Cox multivariate analysis of survival including the DM score and several other predictors of survival as covariates. NKI dataset

Covariate	Odd ratio (95% C.I.)	P-value
DM score (range 0-10)	1.27 (1.13-1.43)	9,84E-005
ER (pos=1, neg=0)	0.56 (0.33-0.96)	0,036
St Gallen (1=low risk, 0-high risk)	0.33 (0.04-2.51)	0,28
LN (positive =1, negative = 0)	0.84 (0.53-1.32)	0,45
NIH (1=low risk, 0-high risk)	0.60 (0.08-4.52)	0,62

Sotiriou-Desmedt dataset

Covariate	Odd ratio (95% C.I.)	P-value
DM score (range 0-10)	1.25 (1.08-1.46)	3,00E-003
Size (cm)	1.23 (0.97-1.57)	0,093
Grade (1-3)	0.79 (0.56-1.09)	0,15
ER (pos=1, neg=0)	1.16 (0.67-2.00)	0,58
Age (years)	1.00 (0.97-1.02)	0,72
LN (positive =1, negative = 0)	1.13 (0.34-3.74)	0,84

Wang dataset
DM score (range 0-10)	1.23 (1.09-1.38)	9,80E-004
ER (pos=1, neg=0)	1.31 (0.81-2.11)	0,27

The authors also performed multivariate Cox analysis to ascertain whether the DM score predicts survival independently of other molecular and histological tumor markers. In all datasets, the DM score is a predictor independent of the available clinical parameters. The results for the Miller dataset, which is the richest in clinical annotation, are reported in Table VII, and the ones for the other datasets in Tables VIII.

Table VII: Multivariate Cox analysis for Miller dataset

Covariate	Odd ratio (95% C.I.)	P-value
LN (positive=1, negative=0)	2.82 (1.53-5.21)	8.95E-04
DM score (range 0-10)	1.32 (1.08-1.60)	0.0057
Size (mm)	1.04 (1.01-1.06)	0.0065
ER (positive=1, negative=0)	3.34 (1.11-10.00)	0.031
Age (years)	1.02 (1.00-1.04)	0.057
PGR (positive=1, negative=0)	0.53 (0.23-1.23))	0.14
P53 (mutant=1, wt=0)	0.97 (0.49-1.95))	0.95
Grade (1-3)	0.99 (0.56-1.75)	0.96

Multivariate Cox analysis for the Miller dataset shows that the DM score is predictive of survival independently of several other predictors.

Table VIII: Number of patients discordantly classified by the three proliferation-based signatures. For each dataset and pair of proliferation-based signatures, the authors report the number of patients classified in different outcome groups, using score cutoffs corresponding to the same sensitivity.

90% sensitivity
	DM	CIN	Proliferation
DM
	0	20 (7.25%)	24 (8.7%)
CIN		0	14 (5.1%)
Proliferation			0

70% sensitivity
	DM	CIN	Proliferation
DM
	0	24 (8.7%)	30 (10.9%)
CIN		0	24 (8.7%)
Proliferation			0

50% sensitivity
	DM	CIN	Proliferation
DM
	0	10 (3.62%)	23 (8.33%)
CIN		0	21 (7.61 %)
Proliferation			0

Multivariate Cox analysis on the Miller dataset using the other proliferation-based signatures gives very similar results, shown in Table IX.

Table IX: Cox multivariate analysis for other proliferation-based signatures. For the CIN and proliferation signature we report the results of the Cox multivariate analysis using the signature score and various other predictors of survival as covariates.

CIN signature
Covariate	Odd ratio (95% C.I.)	P-value
LN	2.86 (1.54-5.29)	8,64E-004
Size	1.04 (1.01-1.06)	5,09E-003
CIN score	1.26 (1.07-1.49)	6,29E-003
ER	3.26 (1.09-9.74)	0,034
Age	1.02 (1.00-1.04)	0,055
PGR	0.54 (0.23-1.26)	0,16
Grade	1.01 (0.57-1.79)	0,98
P53	1.00 (0.51-1.99)	0,99

Proliferation signature
Covariate	Odd ratio (95% C.I.)	P-value
LN	2.78 (1.51-5.15)	1,08E-003
Size	1.04 (1.01-1.07)	5,00E-003
Proliferation score	1.28 (1.07-1.53)	6,65E-003
ER	3.39 (1.13-10.19)	0,03
Age	1.02 (1.00-1.04)	0,072
PGR	0.53 (0.23-1.24)	0,14
P53	0.97 (0.49-1.93)	0,93
Grade	0.98 (0.55-1.75)	0,95

Lymph-node negative patients are a group of particular clinical significance: therefore the authors computed the AUC under ROC curves for the DM signature as a predictor of 5-year survival in the Miller and Sotiriou-Desmedt datasets limited to this subgroup. In both cases the authors find AUC values similar to the ones found for the entire dataset (AUC resp. 0.616 and 0.678). The Wang dataset includes lymph-node negative patients only.

Contribution of specific genes and gene classes to the predictive power of the DM signature.

The authors next asked whether any of the phenotypic class identified by the RNAi screen (chromosome condensation, chromosome integrity, chromosome segregation, spindle assembly and cytokinesis) [6] is especially relevant in separating poor- from good-prognosis patients. The authors computed the contribution of each probeset in the DM signature to the difference in score between poor- and good-outcome patients (see Methods); the authors then compared the contribution of specific gene classes to the total score of the 105 genes of the DM signature. For the three Affymetrix datasets not used as training, the cytokinesis genes (ANLN, CIT, ECT2, KIF23, PRC1, RACGAP1) turned out to contribute, as a group, significantly more than other genes to the difference in score (P-values between 0.0025 and 0.012, two-sided Mann-Whitney U test). The function of these genes is highly conserved, as they are required for cytokinesis in both Drosophila and humans (reviewed in [30]). Interestingly, high z-scores were also observed for ASPM, KIF18A and PLK1 (respectively, 3.99, 3.49 and 3.14). The Drosophila homologues of these genes (asp, Klp67 and polo) play role in multiple mitotic stages and are required for cytokinesis [30]. In addition there is evidence that ASPM and PLK1 are involved in human cell cytokinesis [30]. Thus, it appears that cytokinesis genes have higher prognostic value than other mitotic genes and genes required for chromosome integrity.
In the DM signature, there are a few genes whose reduced expression is negatively correlated with survival (Table II). The gene with the most negative z-score is PIASl (z = - 4.07, averaged on two probesets), an E3 ligase involved in sumoylation of DNA repair proteins including BRCA1 [31]. Remarkably, the expression of this gene is substantially reduced in colon cancers [32].
The authors have shown that the DM signature is highly predictive of survival in five major breast cancer datasets. The DM signature contains two classes of genes required for cell proliferation: genes that maintain the integrity of mitotic chromosomes and genes that mediate mitotic division. Cell proliferation-associated genes have been previously used to construct several unsupervised signatures, and large subsets of this type of genes are included in most supervised signatures [33]. Thus, it has been suggested that genes required for cell proliferation may underlie the prognostic power of many cancer signatures [33].
In agreement with such expectations the authors found that the DM signature has a predictive power for breast cancer outcome similar to two other proliferation-based signatures (the CIN signature [15] and the Proliferation signature of Starmans et al. [11]), and outperforms 4 additional published signatures that contain different proportions of proliferation-related genes, including the supervised 70-gene signature, which is currently used in clinical practice for breast cancer patients [3]. Altogether, these results indicate that signatures enriched in proliferation genes are the most powerful predictors of breast cancer outcome.
High performance of the DM signature may reflect its specifically high content in genes truly involved in cell proliferation. The proliferation-associated genes in the other signatures have been selected on the basis of their periodic expression pattern during the cell cycle and include several genes that, although periodically expressed, are not involved in basic cell cycle processes [10,33]. In contrast, genes underlying either the maintenance of chromosome integrity or mitosis are expected to play essential roles in cell cycle progression and cell proliferation. Thus the DM signature is a strong predictor of survival in breast cancer because it contains a relatively undiluted sample of genes essential for cell proliferation. The expression of these genes should therefore reflect the cell proliferation rate within a cancer better than the gene sets of the other signatures. Consistent with this idea, the authors have shown that most of the DM signature genes with a high predictive power of poor outcome in patients display increased expression (Fig. 4).
The frequency of mitotic cells is one of the criteria used to classify breast cancers in low versus high grade. However, cytological analysis of mitosis proved to be a rather subjective assay with significant inter-observer variations [34]. The analysis of gene expression using the DM signature provides reliable quantitative information on cell proliferation within a breast cancer sample, allowing risk assessments in individual patients.
The authors have shown that a group of genes required for cytokinesis (ANLN, CIT, ECT2, KIF23, PRC1, RACGAP1, ASPM, KJF18A and PLK1) contributes to the predictive power of the DM signature significantly more than the other genes in the signature. All cytokinesis genes display high positive z-scores, indicating that an increased expression level of these genes is negatively correlated with survival. Strikingly, there is evidence that ANLN, ECT2, PRC1, RACGAP1, ASPM, and PLK1 are upregulated in a variety of human cancers and that the overexpression levels of these genes often correlate with poor outcomes in patients (see for example [35-43] and references therein). In addition, it has been shown that two of these cytokinesis genes, ETC2 and ANLN, are amplified in cancer cells [38,44]. These findings raise the questions of why cytokinesis genes have a higher prognostic value and tend to be more upregulated in cancers compared to other mitotic genes. It is possible that overexpression of cytokinesis genes is an oncogenic factor per se. However, the finding that PRC1 overexpression does not result in cell growth enhancement [41] argues against this possibility. Another possibility is that cytokinesis proteins are limited in amount or stability compared to other mitotic proteins. That is, when cell proliferation is strongly enhanced, normal levels of gene transcription and translation would not be sufficient to produce the amounts of cytokinesis proteins required for proper execution of the process. As a result, cancers cell clones overexpressing cytokinesis genes would be favoured over clones in which these genes are normally expressed.
In conclusion, the present invention indicates that the DM signature improves risk stratification for breast cancer patients compared to the major extant signatures. In addition, the identification of new cancer prognostic genes with well-defined biological functions, such as those of the DM signature, provides new prognostic tools based on gene expression. For example, according to a previous approach [6,11,13] the genes of the DM signature could be merged with those of other signatures to further improve risk stratification. Finally, the authors' finding that cytokinesis genes tend to be overexpressed in patients with poor prognosis sets forth this class of genes and their protein products as targets for antimitotic therapies.

References

1.: Dupuy A, Simon RM (2007) J Natl Cancer Inst 99: 147-157.
2.: Wirapati P, et al. (2008) Breast Cancer Res 10: R65.
3.: van't Veer LJ, et al. (2002) Nature 415: 530-536.
4.: van de Vijver MJ, et al. (2002) N Engl J Med 347: 1999-2009.
5.: Chang HY, et al. (2004) PLoS Biol 2: E7.
6.: Chang HY, et al. (2005) Proc Natl Acad Sci USA 102: 3738-3743.
7.: Chi JT, et al. (2006) PLoS Med 3: e47.
8.: Sung FL, et al. (2007) Cancer Lett 253: 74-88.
9.: Winter SC, et al. (2007) Cancer Res 67: 3441-3449.
10.: Whitfield ML, et al. (2002) Mol Biol Cell 13: 1977-2000.
11.: Starmans MH, et al. (2008) Br J Cancer 99: 1884-1890.
12.: Ben-Porath I, et al. (2008) Nat Genet 40: 499-507.
13.: Reyal F, et al. (2008) Breast Cancer Res 10: R93.
14.: Liu R, et al. (2007) N Engl J Med 356: 217-226.
15.: Carter SL, et al. (2006) Nat Genet 38: 1043-1048.
16.: Somma MP, et al. (2008) PLoS Genet 4: e1000126.
17.: Sayers EW, et al. (2010) Nucleic Acids Res 38: D5-16.
18.: Shedden K, et al. (2008) Nat Med 14: 822-827.
19.: Phillips HS, et al. (2006) Cancer Cell 9: 157-173.
20.: Pawitan Y, et al. (2005) Breast Cancer Res 7: R953-964.
21.: Miller LD, et al. (2005) Proc Natl Acad Sci USA 102: 13550-13555.
22.: Wang Y, et al. (2005) Lancet 365: 671-679.
23.: Desmedt C, et al. (2007) Clin Cancer Res 13: 3207-3214.
24.: Sotiriou C, et al. (2006) J Natl Cancer Inst 98: 262-272.
25.: Wilson CL,Miller CJ (2005) Bioinformatics 21: 3683-3685.
26.: Gentleman RC, et al. (2004) Genome Biol 5: R80.
27.: Bild AH, et al. (2006) Nature 439: 353-357.
28.: Freije WA, et al. (2004) CancerRes 64: 6503-6510.
29.: Zhao H, et al. (2006) PLoS Med 3: e13.
30.: Eggert US, Mitchison TJ,Field CM (2006) Annu Rev Biochem 75: 543-566.
31.: Galanty Y, et al. (2009) Nature 462: 935-939.
32.: Coppola D, et al. (2009) J Cancer Res Clin Oncol 135: 1287-1291.
33.: Whitfield ML, George LK, Grant GD,Perou CM (2006) Nat Rev Cancer 6: 99-106.
34.: Paik S, et al. (2004) N Engl J Med 351: 2817-2826.
35.: Suzuki C, et al. (2005) Cancer Res 65: 11314-11325.
36.: Tamura K, et al. (2007) Cancer Res 67: 5117-5125.
37.: Skrzypski M, et al. (2008) Clin Cancer Res 14: 4794-4799.
38.: Fields AP, Justilien V (2009) Adv Enzyme Regul.
39.: Horvath S, et al. (2006) Proc Natl Acad Sci USA 103: 17402-17407.
40.: Lin SY, et al. (2008) Clin Cancer Res 14: 4814-4820.
41.: Shimo A, et al. (2007) Cancer Sci 98: 174-181.
42.: Pellegrino R, et al. (2009) Hepatology .
43.: Schmit TL, et al. (2009) J Invest Dermatol 129: 2843-2853.
44.: Shimizu S, et al. (2007) Oncol Rep 18: 1489-1497.

Claims

A method to predict the mortality risk of a subject (p) affected of breast cancer comprising:
a) measuring the expression level of the genes C15orf44, CASP7, CNOT3, CTPS, CUL4B, CWC15, DCAKD, DDB1, FRGI, MSH6, ORC5L, PCNA, PIAS1, POLA1, PRIM2, PRPF3, RAD54L, RFC2, RPA1, RRM2, SART1, SF3A3, SMCIA, TAF6, TFDP2, TK2, TPR, TYMS, WBP11, WDR46, WDR75, XAB2, XRN2, ZMYM4, MCM3, MCM7, SMC3, NCAPD2, NCAPG, SMC4, SMC2, MASTL, ORC2L, TOP2A, CDT1, BUB3, KNTC1, ZW10, ASCC3L1, CCNB1, CDC40, DHX8, KIAA1310, LSM2, PRPF31, SF3A1, SF3A2, SF3B1, SF3B2, SF3B14, SLU7, SNRPA1, SNRPE, TXNL4A, U2AF1, U2AF2, ANAPC5, ANAPC10, CDC20" KIN, PSMC1, SFRS15. CKAP5, EIF3A, EIF3D, EIF3E, EIF3I, GTF3C3, MAPRE3, NOC3L, RRPIB, TBKI, THOC2, TUBB2C, WDR82, TRRAP, TUBGCP4, TUBG2, ASPM, CENPJ, MKI67IP, PPP1R8, CDC2, KIFC1, KJF11, KIF18A, AURKC, RBBP7, PLK1, ECT2, KIF23, PRC1, RACGAP1, ANLN, CIT in a biological sample, obtaining the prognostic score, S(p), that indicates the expression levels of said genes in said subject (p) affected of cancer, and

b) predicting the mortality risk of said subject (p) affected of cancer comparing said prognostic score, S(p), to a cut off value (cut off threshold).
The method according to claim 1 wherein the expression level of said genes is measured by means of quantitative detection of the transcript sequences selected from the group SEQ ID No 1 to SEQ ID No. 217.
The method according to any one of previous claim wherein the expression level of said genes is detected by means of microarray.
The method according to any one of previous claim wherein the biological sample is selected from the group of: blood, tumour cell, frozen or fixed tissue sections, biopsy, biological fluid.
The method according to any one of previous claim wherein the mortality risk is assigned as follows:
iv) to the class "low risk" if the prognostic score, S(p), is lower than the cut off threshold, or

v) to the class "high risk" if the prognostic score, S(p), is greater than the cut off threshold, and optionally

vi) to the class "intermediate" if the prognostic score, S(p), is between two cut off threshold values.
The method according to any one of previous claim wherein the prognostic score, S(p), is calculated according to the following formula: $S (p) = \sum_{g} x (g p) z (g)$

wherein
x(g,p) is the expression level expressed in logarithmic base 2 of the probeset g in the patient p;
z(g) is the z-score of the probeset g calculated in the Pawitan dataset;
wherein the probeset g comprises a group of 217 probes, each one being specific and selective for one of the gene transcript belonging to the group of SEQ ID No. 1 to SEQ ID No. 217.
The method according to claim 6 wherein the z-score for each probe is the one calculated in the Pawitan database reported in table II.
A kit to detect the transcript expression level of genes C15orf44, CASP7, CNOT3, CTPS, CUL4B, CWC15, DCAKD, DDB1, FRG1, MSH6, ORC5L, PCNA, PIAS1, POLA1, PRIM2, PRPF3, RAD54L, RFC2, RPA1, RRM2, SART1, SF3A3, SMC1A, TAF6, TFDP2, TK2, TPR, TYMS, WBP11, WDR46, WDR75, XAB2, XRN2, ZMYM4, MCM3, MCM7, SMC3, NCAPD2, NCAPG, SMC4, SMC2, MASTL, ORC2L, TOP2A, CDT1, BUB3, KNTC1, ZW10, ASCC3L1, CCNB1, CDC40, DHX8, KIAA1310, LSM2, PRPF31, SF3A1, SF3A2, SF3B1, SF3B2, SF3B14, SLU7, SNRPA1, SNRPE, TXNL4A, U2AF1, U2AF2, ANAPC5, ANAPC10, CDC20" KIN, PSMC1, SFRS15. CKAP5, EIF3A, EIF3D, EIF3E, EIF3I, GTF3C3, MAPRE3, NOC3L, RRPIB, TBKI, THOC2, TUBB2C, WDR82, TRRAP, TUBGCP4, TUBG2, ASPM, CENPJ, MK167IP, PPP1R8, CDC2, KIFC1, KIF11, KIF18A, AURKC, RBBP7, PLK1, ECT2, KIF23, PRC1, RACGAP1, ANLN, CIT,
comprising:
- for each of said genes, sequence specific amplification means to obtain amplified nucleic acids having sequences comprised in the transcribed region thereof;

- quantitative detection means of said amplified nucleic acids;

- appropriate reagents.
The kit according to claim 8 wherein said amplified nucleic acids consist of :
for C15orf44, SEQ ID No. 145; for CASP7, SEQ ID No. 189; for CNOT3, SEQ ID No. 66 and/or SEQ ID No. 138 and/or SEQ ID No.167; for CTPS, SEQ ID No. 39; for CUL4B, SEQ ID No. 113 and/or SEQ ID No. 152 and/or SEQ ID No. 165 and/or SEQ ID No. 212; for CWC15, SEQ ID No. 159; for DCAKD, SEQ ID No. 126 and/or SEQ ID No. 140 and/or SEQ ID No. 190; for DDB1, SEQ ID No. 38; for FRG1, SEQ ID No. 195; for MSH6, SEQ ID No. 46 and/or SEQ ID No. 61 and /or SEQ ID No. 153 and/or SEQ ID No. 187; for ORC5L, SEQ ID No. 70 and/or SEQ ID No. 79 and/or SEQ ID No. 109; for PCNA, SEQ ID No. 51; for PIAS1, SEQ ID No. 211 and/or SEQ ID No. 216 and/or SEQ ID No. 217; for POLA1, SEQ ID No. 147; for PRIM2, SEQ ID No. 43 and/or SEQ ID No. 56 and/or SEQ ID No. 88; for PRPF3, SEQ ID No. 170; for RAD54L, SEQ ID No. 75; for RFC2, SEQ ID No. 42 and/or SEQ ID No. 48; for RPA1, SEQ ID No. 64 and/or SEQ ID No. 103; for RRM2, SEQ ID No. 3 and/or SEQ ID No. 9; for SART1, SEQ ID No. 124; for SF3A3, SEQ ID No. 201; for SMC1A, SEQ ID No. 115 and/or SEQ ID No. 179 and/or SEQ ID No. 207; for TAF6, SEQ ID No. 68; for TFDP2, SEQ ID No. 86 and/or SEQ ID No. 118 and/or SEQ ID No. 210; for TK2, SEQ ID No. 37 and/or SEQ ID No. 156 and/or SEQ ID No. 171 and/or SEQ ID No. 172; for TPR, SEQ ID No. 99 and/or SEQ ID No. 108 and/or SEQ ID No. 182 and/or SEQ ID No. 204; for TYMS, SEQ ID No. 32 and/or SEQ ID No. 125; for WBP11, SEQ ID No. 65 and/or SEQ ID No. 67; for WDR46, SEQ ID No. 93; for WDR75, SEQ ID No. 158; for XAB2, SEQ ID No. 180; for XRN2, SEQ ID No. 81 and/or SEQ ID No. 84; for ZMYM4, SEQ ID No. 192 and/or SEQ ID No. 196 and/or SEQ ID No. 213; for MCM3, SEQ ID No. 34; for MCM7, SEQ ID No. 28 and/or SEQ ID No. 52; for SMC3, SEQ ID No. 185 and/or SEQ ID No. 193 and/or SEQ ID No. 209; for NCAPD2, SEQ ID No. 106; for NCAPG, SEQ ID No.22 and/or SEQ ID No. 24; for SMC4, SEQ ID No. 33 and/or SEQ ID No. 54 and/or SEQ ID No. 141; for SMC2, SEQ ID No. 45 and/or SEQ ID No. 127; for MASTL, SEQ ID No. 11; for ORC2L, SEQ ID No. 104; for TOP2A, SEQ ID No. 20 and/or SEQ ID No. 62 and/or SEQ ID No. 96; for CDT1, SEQ ID No. 2 and/or SEQ ID No. 36; for BUB3, SEQ ID No. 57 and/or SEQ ID No. 139 and/or SEQ ID No. 148 and/or SEQ ID No. 174 and/or SEQ ID No. 178; for KNTC1, SEQ ID No. 35; for ZW10, SEQ ID No. 143; for ASCC3L1, SEQ ID No. 55 and/or SEQ ID No. 135 and/or SEQ ID No. 150; for CCNB1, SEQ ID No. 7 and/or SEQ ID No. 14; for CDC40, SEQ ID No. 100 and/or SEQ ID No. 177; for DHX8, SEQ ID No. 58 and/or SEQ ID No. 120 and/or SEQ ID No. 121; for KIAA1310, SEQ ID No. 160 and/or SEQ ID No. 183 and/or SEQ ID No. 188; for LSM2, SEQ ID No. 137; for PRPF31, SEQ ID No. 60 and/or SEQ ID No. 91 and/or SEQ ID No. 184; for SF3A1, SEQ ID No. 98 and/or SEQ ID No. 119 and/or SEQ ID No. 162 and/or SEQ ID No. 173; for SF3A2, SEQ ID No. 169 and/or SEQ ID No. 176; for SF3B1, SEQ ID No. 194 and/or SEQ ID No. 203 and/or SEQ ID No. 208 and/or SEQ ID No. 214; for SF3B2, SEQ ID No. 77; for SF3B14, SEQ ID No. 10; for SLU7, SEQ ID No. 149 and/or SEQ ID No. 151; for SNRPA1, SEQ ID No. 23 and/or SEQ ID No. 49 and/or SEQ ID No. 71 and/or SEQ ID No. 181; for SNRPE, SEQ ID No. 72 and/or SEQ ID No. 136; for TXNL4A, SEQ ID No. 26 and/or SEQ ID No. 134; for U2AF1, SEQ ID No. 30 and/or SEQ ID No. 82 and/or SEQ ID No. 102 and/or SEQ ID No. 131; for U2AF2, SEQ ID No. 94 and/or SEQ ID No. 146 and/or SEQ ID No. 155 and/or SEQ ID No. 161; for ANAPC5, SEQ ID No. 85 and/or SEQ ID No. 95 and/or SEQ ID No. 97 and/or SEQ ID No. 112 and/or SEQ ID No. 117; for ANAPC10, SEQ ID No. 129; for CDC20, SEQ ID No. 17; for KIN, SEQ ID No. 111 and/or SEQ ID No. 144; for PSMC1, SEQ ID No. 25; for SFRS15, SEQ ID No. 50 and/or SEQ ID No. 63 and/or SEQ ID No. 80 and/or SEQ ID No. 142 and/or SEQ ID No. 197; for CKAP5, SEQ ID No. 21; for EIF3A, SEQ ID No. 175 and/or SEQ ID No. 186 and/or SEQ ID No. 202; for EIF3D, SEQ ID No. 101; for EIF3E, SEQ ID No. 154; for EIF31, SEQ ID No. 114; for GTF3C3, SEQ ID No. 74 and/or SEQ ID No. 163; for MAPRE3, SEQ ID No. 116 and/or SEQ ID No. 128 and/or SEQ ID No. 130 and/or SEQ ID No. 133; for NOC3L, SEQ ID No. 164; for RRPIB, SEQ ID No. 105 and/or SEQ ID No. 123; for TBKI, SEQ ID No. 198; for THOC2, SEQ ID No. 110 and/or SEQ ID No. 132 and/or SEQ ID No. 199 and/or SEQ ID No. 205; for TUBB2C, SEQ ID No. 4 and/or SEQ ID No. 5; for WDR82, SEQ ID No. 191; for TRRAP, SEQ ID No. 69 and/or SEQ ID No. 73; for TUBGCP4, SEQ ID No. 76 and/or SEQ ID No. 215; for TUBG2, SEQ ID No. 157; for ASPM, SEQ ID No. 6 and/or SEQ ID No. 47 and/or SEQ ID No. 53; for CENPJ, SEQ ID No. 87 and/or SEQ ID No. 92 and/or SEQ ID No. 107; for MK167IP, SEQ ID No. 41 and/or SEQ ID No. 89 and/or SEQ ID No. 200; for PPPIR8, SEQ ID No. 168; for CDC2, SEQ ID No. 15 and/or SEQ ID No. 16 and/or SEQ ID No. 31 and/or SEQ ID No. 206; for KIFC1, SEQ ID No. 19; for KIF11, SEQ ID No. 29; for KIF18A, SEQ ID No. 18; for AURKC, SEQ ID No. 90; for RBBP7, SEQ ID No. 166; for PLK1, SEQ ID No. 27; for ECT2, SEQ ID No. 40 and/or SEQ ID No. 59 and/or SEQ ID No. 83; for KIF23, SEQ ID No. 8 and/or SEQ ID No. 44; for PRC1, SEQ ID No. 13; for RACGAP1, SEQ ID No. 12; for ANLN, SEQ ID No. 1; for CIT, SEQ ID No. 78 and/or SEQ ID No. 122.
The kit according to claim 8 or 9 further comprising sequence specific amplification means to obtain amplified nucleic acids having sequences comprised in the transcribed region of genes H3F3A and/or PPAN-P2RY11 and/or KIF4.
A microarray consisting of:
a) solid supporting means, and

b) for each of the genes C15orf44, CASP7, CNOT3, CTPS, CUL4B, CWC15, DCAKD, DDB1, FRG1, MSH6, ORC5L, PCNA, PIAS1, POLA1, PRIM2, PRPF3, RAD54L, RFC2, RPA1, RRM2, SART1, SF3A3, SMCIA, TAF6, TFDP2, TK2, TPR, TYMS, WBP11, WDR46, WDR75, XAB2, XRN2, ZMYM4, MCM3, MCM7, SMC3, NCAPD2, NCAPG, SMC4, SMC2, MASTL, ORC2L, TOP2A, CDT1, BUB3, KNTC1, ZW10, ASCC3L1, CCNB1, CDC40, DHX8, KIAA1310, LSM2, PRPF31, SF3A1, SF3A2, SF3B1, SF3B2, SF3B14, SL U7, SNRPA1, SNRPE, TXNL4A, U2AF1, U2AF2, ANAPC5, ANAPC10, CDC20" KIN, PSMC1, SFRS15. CKAP5, EIF3A, EIF3D, EIF3E, EIF3I, GTF3C3, MAPRE3, NOC3L, RRPIB, TBKI, THOC2, TUBB2C, WDR82, TRRAP, TUBGCP4, TUBG2, ASPM, CENPJ, MKI67IP, PPPIR8, CDC2, KIFC1, KJF11, KJF18A, AURKC, RBBP7, PLK1, ECT2, KIF23, PRC1, RACGAP1, ANLN, CIT, at least one oligonucleotide able to specifically hybridize to a sequence comprised in the transcribed region thereof.
The microarray according to claim 11 wherein the sequences comprised in the transcribed region of said genes consist of : for C15orf44, SEQ ID No. 145; for CASP7, SEQ ID No. 189; for CNOT3, SEQ ID No. 66 and/or SEQ ID No. 138 and/or SEQ ID No.167; for CTPS, SEQ ID No. 39; for CUL4B, SEQ ID No. 113 and/or SEQ ID No. 152 and/or SEQ ID No. 165 and/or SEQ ID No. 212; for CWC15, SEQ ID No. 159; for DCAKD, SEQ ID No. 126 and/or SEQ ID No. 140 and/or SEQ ID No. 190; for DDB1, SEQ ID No. 38; for FRG1, SEQ ID No. 195; for MSH6, SEQ ID No. 46 and/or SEQ ID No. 61 and /or SEQ ID No. 153 and/or SEQ ID No. 187; for ORC5L, SEQ ID No. 70 and/or SEQ ID No. 79 and/or SEQ ID No. 109; for PCNA, SEQ ID No. 51; for PIAS1, SEQ ID No. 211 and/or SEQ ID No. 216 and/or SEQ ID No. 217; for POLA1, SEQ ID No. 147; for PRIM2, SEQ ID No. 43 and/or SEQ ID No. 56 and/or SEQ ID No. 88; for PRPF3, SEQ ID No. 170; for RAD54L, SEQ ID No. 75; for RFC2, SEQ ID No. 42 and/or SEQ ID No. 48; for RPA1, SEQ ID No. 64 and/or SEQ ID No. 103; for RRM2, SEQ ID No. 3 and/or SEQ ID No. 9; for SART1, SEQ ID No. 124; for SF3A3, SEQ ID No. 201; for SMCIA, SEQ ID No. 115 and/or SEQ ID No. 179 and/or SEQ ID No. 207; for TAF6, SEQ ID No. 68; for TFDP2, SEQ ID No. 86 and/or SEQ ID No. 118 and/or SEQ ID No. 210; for TK2, SEQ ID No. 37 and/or SEQ ID No. 156 and/or SEQ ID No. 171 and/or SEQ ID No. 172; for TPR, SEQ ID No. 99 and/or SEQ ID No. 108 and/or SEQ ID No. 182 and/or SEQ ID No. 204; for TYMS, SEQ ID No. 32 and/or SEQ ID No. 125; for WBP11, SEQ ID No. 65 and/or SEQ ID No. 67; for WDR46, SEQ ID No. 93; for WDR75, SEQ ID No. 158; for XAB2, SEQ ID No. 180; for XRN2, SEQ ID No. 81 and/or SEQ ID No. 84; for ZMYM4, SEQ ID No. 192 and/or SEQ ID No. 196 and/or SEQ ID No. 213; for MCM3, SEQ ID No. 34; for MCM7, SEQ ID No. 28 and/or SEQ ID No. 52; for SMC3, SEQ ID No. 185 and/or SEQ ID No. 193 and/or SEQ ID No. 209; for NCAPD2, SEQ ID No. 106; for NCAPG, SEQ ID No.22 and/or SEQ ID No. 24; for SMC4, SEQ ID No. 33 and/or SEQ ID No. 54 and/or SEQ ID No. 141; for SMC2, SEQ ID No. 45 and/or SEQ ID No. 127; for MASTL, SEQ ID No. 11; for ORC2L, SEQ ID No. 104; for TOP2A, SEQ ID No. 20 and/or SEQ ID No. 62 and/or SEQ ID No. 96; for CDT1, SEQ ID No. 2 and/or SEQ ID No. 36; for BUB3, SEQ ID No. 57 and/or SEQ ID No. 139 and/or SEQ ID No. 148 and/or SEQ ID No. 174 and/or SEQ ID No. 178; for KNTC1, SEQ ID No. 35; for ZW10, SEQ ID No. 143; for ASCC3L1, SEQ ID No. 55 and/or SEQ ID No. 135 and/or SEQ ID No. 150; for CCNB1, SEQ ID No. 7 and/or SEQ ID No. 14; for CDC40, SEQ ID No. 100 and/or SEQ ID No. 177; for DHX8, SEQ ID No. 58 and/or SEQ ID No. 120 and/or SEQ ID No. 121; for KIAA1310, SEQ ID No. 160 and/or SEQ ID No. 183 and/or SEQ ID No. 188; for LSM2, SEQ ID No. 137; for PRPF31, SEQ ID No. 60 and/or SEQ ID No. 91 and/or SEQ ID No. 184; for SF3A1, SEQ ID No. 98 and/or SEQ ID No. 119 and/or SEQ ID No. 162 and/or SEQ ID No. 173; for SF3A2, SEQ ID No. 169 and/or SEQ ID No. 176; for SF3B1, SEQ ID No. 194 and/or SEQ ID No. 203 and/or SEQ ID No. 208 and/or SEQ ID No. 214; for SF3B2, SEQ ID No. 77; for SF3B14, SEQ ID No. 10; for SLU7, SEQ ID No. 149 and/or SEQ ID No. 151; for SNRPA1, SEQ ID No. 23 and/or SEQ ID No. 49 and/or SEQ ID No. 71 and/or SEQ ID No. 181; for SNRPE, SEQ ID No. 72 and/or SEQ ID No. 136; for TXNL4A, SEQ ID No. 26 and/or SEQ ID No. 134; for U2AF1, SEQ ID No. 30 and/or SEQ ID No. 82 and/or SEQ ID No. 102 and/or SEQ ID No. 131; for U2AF2, SEQ ID No. 94 and/or SEQ ID No. 146 and/or SEQ ID No. 155 and/or SEQ ID No. 161; for ANAPC5, SEQ ID No. 85 and/or SEQ ID No. 95 and/or SEQ ID No. 97 and/or SEQ ID No. 112 and/or SEQ ID No. 117; for ANAPC10, SEQ ID No. 129; for CDC20, SEQ ID No. 17; for KIN, SEQ ID No. 111 and/or SEQ ID No. 144; for PSMC1, SEQ ID No. 25; for SFRS15, SEQ ID No. 50 and/or SEQ ID No. 63 and/or SEQ ID No. 80 and/or SEQ ID No. 142 and/or SEQ ID No. 197; for CKAP5, SEQ ID No. 21; for EIF3A, SEQ ID No. 175 and/or SEQ ID No. 186 and/or SEQ ID No. 202; for EIF3D, SEQ ID No. 101; for EIF3E, SEQ ID No. 154; for EIF3I, SEQ ID No. 114; for GTF3C3, SEQ ID No. 74 and/or SEQ ID No. 163; for MAPRE3, SEQ ID No. 116 and/or SEQ ID No. 128 and/or SEQ ID No. 130 and/or SEQ ID No. 133; for NOC3L, SEQ ID No. 164; for RRP1B, SEQ ID No. 105 and/or SEQ ID No. 123; for TBK1, SEQ ID No. 198; for THOC2, SEQ ID No. 110 and/or SEQ ID No. 132 and/or SEQ ID No. 199 and/or SEQ ID No. 205; for TUBB2C, SEQ ID No. 4 and/or SEQ ID No. 5; for WDR82, SEQ ID No. 191; for TRRAP, SEQ ID No. 69 and/or SEQ ID No. 73; for TUBGCP4, SEQ ID No. 76 and/or SEQ ID No. 215; for TUBG2, SEQ ID No. 157; for ASPM, SEQ ID No. 6 and/or SEQ ID No. 47 and/or SEQ ID No. 53; for CENPJ, SEQ ID No. 87 and/or SEQ ID No. 92 and/or SEQ ID No. 107; for MKI67IP, SEQ ID No. 41 and/or SEQ ID No. 89 and/or SEQ ID No. 200; for PPP1R8, SEQ ID No. 168; for CDC2, SEQ ID No. 15 and/or SEQ ID No. 16 and/or SEQ ID No. 31 and/or SEQ ID No. 206; for KIFC1, SEQ ID No. 19; for KIF11, SEQ ID No. 29; for KIF18A, SEQ ID No. 18; for AURKC, SEQ ID No. 90; for RBBP7, SEQ ID No. 166; for PLK1, SEQ ID No. 27; for ECT2, SEQ ID No. 40 and/or SEQ ID No. 59 and/or SEQ ID No. 83; for KIF23, SEQ ID No. 8 and/or SEQ ID No. 44; for PRC1, SEQ ID No. 13; for RACGAP1, SEQ ID No. 12; for ANLN, SEQ ID No. 1; for CIT, SEQ ID No. 78 and/or SEQ ID No. 122.
The microarray according to claim 11 or 12 further comprising at least one oligonucleotide able to specifically hybridize to a sequence comprised in the transcribed region of genes H3F3A and/or PPAN-P2RY11 and/or KIF4.